2 packs: 15% off with code BUY2. 3 or more: 20% off with code BUY3.

Free Google Professional Cloud Security Engineer Practice Questions

94 free Google Professional Cloud Security Engineer practice questions from the CertStash pack of 376, each with the correct answer and a full explanation. The rule is the same for every exam: ten free on the product page, 94 free here, all 376 in the pack, and each step includes the one before it.

Question 1

Your team needs to make sure that a Compute Engine instance does not have access to the internet or to any Google APIs or services.

Which two settings must remain disabled to meet these requirements? (Choose two.)

  1. Public IP
  2. IP Forwarding
  3. Private Google Access
  4. Static routes
  5. IAM Network User Role
Show answer and explanation

Correct answer: A, C. Public IP; Private Google Access

To prevent a Compute Engine instance from accessing the internet or Google APIs, you must disable Public IP (which provides external internet connectivity) and Private Google Access (which allows private access to Google APIs without a public IP). Without a public IP, the instance cannot reach the internet directly. Without Private Google Access enabled, even instances with only private IPs cannot reach Google services. IP Forwarding is for routing between networks, Static routes define how traffic is routed, and IAM Network User Role controls who can use networks, none of these directly control API or internet access for the instance itself.

Why the other options are wrong

  • B. IP Forwarding allows the instance to route traffic between networks but doesn't prevent internet or API access.
  • D. Static routes define routing paths but don't inherently block internet or API access.
  • E. IAM Network User Role is an identity permission that doesn't control network connectivity.

Question 2

Which two implied firewall rules are defined on a VPC network? (Choose two.)

  1. A rule that allows all outbound connections
  2. A rule that denies all inbound connections
  3. A rule that blocks all inbound port 25 connections
  4. A rule that blocks all outbound connections
  5. A rule that allows all inbound port 80 connections
Show answer and explanation

Correct answer: A, B. A rule that allows all outbound connections; A rule that denies all inbound connections

Every VPC network has two implied firewall rules: one that allows all outbound connections by default, and one that denies all inbound connections by default. These baseline rules ensure that instances can initiate outbound traffic while protecting against unsolicited inbound traffic. Administrators must explicitly create firewall rules to allow specific inbound traffic. There are no default rules that specifically block port 25, block all outbound traffic, or allow inbound port 80.

Why the other options are wrong

  • C. There is no implied firewall rule blocking port 25; this would require an explicit rule.
  • D. Outbound connections are allowed by default, not blocked.
  • E. There is no implied rule allowing inbound port 80; all inbound traffic is denied by default.

Question 3

A customer needs an alternative to storing their plain text secrets in their source-code management (SCM) system.

How should the customer achieve this using Google Cloud Platform?

  1. Use Cloud Source Repositories, and store secrets in Cloud SQL.
  2. Encrypt the secrets with a Customer-Managed Encryption Key (CMEK), and store them in Cloud Storage.
  3. Run the Cloud Data Loss Prevention API to scan the secrets, and store them in Cloud SQL.
  4. Deploy the SCM to a Compute Engine VM with local SSDs, and enable preemptible VMs.
Show answer and explanation

Correct answer: B. Encrypt the secrets with a Customer-Managed Encryption Key (CMEK), and store them in Cloud Storage.

The best practice for storing secrets outside of source code is to encrypt them with a Customer-Managed Encryption Key (CMEK) and store them in Cloud Storage. This approach keeps secrets out of the SCM system while providing encryption at rest and access control. Cloud Source Repositories doesn't solve the secret storage problem. The Data Loss Prevention API scans for secrets but doesn't manage their storage. Deploying SCM to a Compute Engine VM replicates the on-premises problem and doesn't address secure secret management.

Why the other options are wrong

  • A. Cloud Source Repositories is a version control system and doesn't provide secure secret storage; storing secrets in Cloud SQL alongside code doesn't solve the SCM problem.
  • C. The Data Loss Prevention API detects secrets but doesn't provide a secure storage solution.
  • D. Running SCM on Compute Engine with local SSDs doesn't provide encryption or centralized secret management and replicates the original problem.

Question 4

Your team wants to centrally manage GCP IAM permissions from their on-premises Active Directory Service. Your team wants to manage permissions by AD group membership.

What should your team do to meet these requirements?

  1. Set up Cloud Directory Sync to sync groups, and set IAM permissions on the groups.
  2. Set up SAML 2.0 Single Sign-On (SSO), and assign IAM permissions to the groups.
  3. Use the Cloud Identity and Access Management API to create groups and IAM permissions from Active Directory.
  4. Use the Admin SDK to create groups and assign IAM permissions from Active Directory.
Show answer and explanation

Correct answer: A. Set up Cloud Directory Sync to sync groups, and set IAM permissions on the groups.

Cloud Directory Sync synchronizes Active Directory groups to Google Cloud Identity, allowing teams to manage GCP IAM permissions based on AD group membership. This provides centralized management of both identity and access control. SAML 2.0 SSO handles authentication but not group-based permission management in the way required. The Cloud Identity and Access Management API and Admin SDK require programmatic integration and don't provide the automatic synchronization that Cloud Directory Sync offers for group-based management.

Why the other options are wrong

  • B. SAML 2.0 SSO enables single sign-on but doesn't centrally manage group-based IAM permissions from Active Directory.
  • C. The IAM API requires manual integration and doesn't automatically sync AD groups for permission management.
  • D. The Admin SDK also requires manual integration and doesn't provide automatic synchronization of AD groups.

Question 5

When creating a secure container image, which two items should you incorporate into the build if possible? (Choose two.)

  1. Ensure that the app does not run as PID 1.
  2. Package a single app as a container.
  3. Remove any unnecessary tools not needed by the app.
  4. Use public container images as a base image for the app.
  5. Use many container image layers to hide sensitive information.
Show answer and explanation

Correct answer: B, C. Package a single app as a container; Remove any unnecessary tools not needed by the app.

When building secure container images, you should package a single app per container to minimize attack surface and simplify updates, and remove any unnecessary tools not needed by the app to reduce vulnerabilities and image size. Running as PID 1 is actually a standard practice for containerized applications. Public base images should be avoided in favor of minimal trusted bases. Using many layers to hide information provides false security through obscurity and is not a security best practice.

Why the other options are wrong

  • A. Running as PID 1 is the standard and expected pattern for containerized applications.
  • D. Public container images may contain vulnerabilities; using verified minimal base images or building from scratch is more secure.
  • E. Using multiple layers to hide information is security through obscurity and provides no actual protection; container layers are inspectable.

Question 6

A customer needs to launch a 3-tier internal web application on Google Cloud Platform (GCP). The customer's internal compliance requirements dictate that end-user access may only be allowed if the traffic seems to originate from a specific known good CIDR. The customer accepts the risk that their application will only have SYN flood DDoS protection. They want to use GCP's native SYN flood protection.

Which product should be used to meet these requirements?

  1. Cloud Armor
  2. VPC Firewall Rules
  3. Cloud Identity and Access Management
  4. Cloud CDN
Show answer and explanation

Correct answer: A. Cloud Armor

Cloud Armor is Google Cloud's WAF and DDoS protection service that allows you to define security policies based on source IP ranges (CIDR blocks) and provides native SYN flood protection. It can restrict access to only traffic from known good CIDRs while providing the SYN flood DDoS protection the customer requires. VPC Firewall Rules operate at the network layer but don't provide DDoS protection; they would require complex rules and don't include SYN flood mitigation. Cloud Identity and Access Management controls who can access GCP resources, not application traffic. Cloud CDN is for content delivery, not access control or DDoS protection.

Why the other options are wrong

  • B. VPC Firewall Rules control network traffic but don't provide DDoS protection or native SYN flood protection.
  • C. Cloud IAM manages user access to GCP resources, not incoming application traffic filtering.
  • D. Cloud CDN is for content delivery optimization, not for DDoS protection or CIDR-based access control.

Question 7

A company is running workloads in a dedicated server room. They must only be accessed from within the private company network. You need to connect to these workloads from Compute Engine instances within a Google Cloud Platform project.

Which two approaches can you take to meet the requirements? (Choose two.)

  1. Configure the project with Cloud VPN.
  2. Configure the project with Shared VPC.
  3. Configure the project with Cloud Interconnect.
  4. Configure the project with VPC peering.
  5. Configure all Compute Engine instances with Private Access.
Show answer and explanation

Correct answer: A, C. Configure the project with Cloud VPN; Configure the project with Cloud Interconnect.

To connect Compute Engine instances to on-premises dedicated servers while keeping traffic private, you can use Cloud VPN to create an encrypted tunnel between the GCP network and the on-premises network, or Cloud Interconnect for a dedicated private connection. Both provide secure connectivity to private networks. Shared VPC is for sharing networks across GCP projects, not connecting to on-premises. VPC peering connects separate GCP VPC networks, not on-premises infrastructure. Private Access enables Google APIs access but doesn't connect to on-premises workloads.

Why the other options are wrong

  • B. Shared VPC shares networks across GCP projects, not connecting to on-premises infrastructure.
  • D. VPC peering connects separate Google Cloud VPC networks to each other, not to on-premises networks.
  • E. Private Access (Private Google Access) enables access to Google APIs privately but doesn't facilitate on-premises connectivity.

Question 8

A customer implements Cloud Identity-Aware Proxy for their ERP system hosted on Compute Engine. Their security team wants to add a security layer so that the ERP systems only accept traffic from Cloud Identity-Aware Proxy.

What should the customer do to meet these requirements?

  1. Make sure that the ERP system can validate the JWT assertion in the HTTP requests.
  2. Make sure that the ERP system can validate the identity headers in the HTTP requests.
  3. Make sure that the ERP system can validate the x-forwarded-for headers in the HTTP requests.
  4. Make sure that the ERP system can validate the user's unique identifier headers in the HTTP requests.
Show answer and explanation

Correct answer: A. Make sure that the ERP system can validate the JWT assertion in the HTTP requests.

Cloud Identity-Aware Proxy adds a JWT (JSON Web Token) assertion to HTTP requests that the backend application can validate to ensure traffic originates from IAP. The ERP system should validate the JWT assertion in the X-Goog-IAP-JWT-Assertion header to confirm the request passed through IAP authentication. Identity headers and x-forwarded-for headers are not the standard IAP validation mechanism. While IAP can provide user identifier information, JWT validation is the primary security mechanism for ensuring traffic originated from IAP.

Why the other options are wrong

  • B. Identity headers are not the standard IAP validation method; JWT assertion validation is required.
  • C. X-forwarded-for headers indicate proxy forwarding but don't provide authentication or proof that traffic passed through IAP.
  • D. While user identifiers may be available, JWT assertion validation is the required security mechanism to verify IAP authorization.

Question 9

A company has been running their application on Compute Engine. A bug in the application allowed a malicious user to repeatedly execute a script that results in the Compute Engine instance crashing. Although the bug has been fixed, you want to get notified in case this hack re-occurs.

What should you do?

  1. Create an Alerting Policy in Stackdriver using a Process Health condition, checking that the number of executions of the script remains below the desired threshold. Enable notifications.
  2. Create an Alerting Policy in Stackdriver using the CPU usage metric. Set the threshold to 80% to be notified when the CPU usage goes above this 80%.
  3. Log every execution of the script to Stackdriver Logging. Create a User-defined metric in Stackdriver Logging on the logs, and create a Stackdriver Dashboard displaying the metric.
  4. Log every execution of the script to Stackdriver Logging. Configure BigQuery as a log sink, and create a BigQuery scheduled query to count the number of executions in a specific timeframe.
Show answer and explanation

Correct answer: A. Create an Alerting Policy in Stackdriver using a Process Health condition, checking that the number of executions of the script remains below the desired threshold. Enable notifications.

When a malicious script execution causes crashes, you need to detect abnormal process behavior. A Process Health condition in Stackdriver Alerting directly monitors process execution counts and can trigger notifications when the number of script executions exceeds a threshold. This provides real-time alerts on the specific suspicious behavior. Option B uses CPU as a proxy metric which is too indirect and generic. Options C and D focus on logging and analysis rather than real-time alerting for immediate notification.

Why the other options are wrong

  • B. CPU usage at 80% is a generic threshold unrelated to the specific script execution attack pattern.
  • C. Creating a user-defined metric and dashboard provides visibility but not automated alerting for immediate notification.
  • D. BigQuery scheduled queries provide periodic analysis but lack real-time alerting for immediate response to the attack.

Question 10

Your team needs to obtain a unified log view of all development cloud projects in your SIEM. The development projects are under the NONPROD organization folder with the test and pre-production projects. The development projects share the ABC-BILLING billing account with the rest of the organization.

Which logging export strategy should you use to meet the requirements?

  1. 1. Export logs to a Cloud Pub/Sub topic with folders/NONPROD parent and includeChildren property set to True in a dedicated SIEM project. 2. Subscribe SIEM to the topic.
  2. 1. Create a Cloud Storage sink with billingAccounts/ABC-BILLING parent and includeChildren property set to False in a dedicated SIEM project. 2. Process Cloud Storage objects in SIEM.
  3. 1. Export logs in each dev project to a Cloud Pub/Sub topic in a dedicated SIEM project. 2. Subscribe SIEM to the topic.
  4. 1. Create a Cloud Storage sink with a publicly shared Cloud Storage bucket in each project. 2. Process Cloud Storage objects in SIEM.
Show answer and explanation

Correct answer: A. 1. Export logs to a Cloud Pub/Sub topic with folders/NONPROD parent and includeChildren property set to True in a dedicated SIEM project. 2. Subscribe SIEM to the topic.

To obtain unified logs from all development projects under the NONPROD folder with minimal configuration, you should create a single log sink at the folder level with includeChildren enabled. This automatically captures logs from all projects within that folder without needing individual sink configurations per project. Cloud Pub/Sub is appropriate for real-time streaming to a SIEM system. A folder-level sink with includeChildren=True is the most efficient approach for unified log collection across multiple related projects.

Why the other options are wrong

  • B. Using billingAccounts/ABC-BILLING parent includes all projects under that billing account (including test and pre-production), which is broader than just the NONPROD folder's development projects.
  • C. Exporting from each development project individually requires multiple sink configurations and lacks the unified approach offered by folder-level exports.
  • D. Publicly shared Cloud Storage buckets violate security principles and requiring per-project configuration is operationally inefficient.

What people say about the full Professional Cloud Security Engineer pack 5.0 from 1

The main thing is that every topic was explained in depth, but without unnecessary details which i like the most. Great Job!
Farah AzizProfessional Cloud Security Engineer

Question 11

You are working with a client that is concerned about control of their encryption keys for sensitive data. The client does not want to store encryption keys at rest in the same cloud service provider (CSP) as the data that the keys are encrypting.

Which Google Cloud encryption solutions should you recommend to this client?

(Choose two.)

  1. Customer-supplied encryption keys.
  2. Google default encryption
  3. Secret Manager
  4. Cloud External Key Manager
  5. Customer-managed encryption keys
Show answer and explanation

Correct answer: A, D. Customer-supplied encryption keys; Cloud External Key Manager

Customer-supplied encryption keys allow clients to provide their own keys that are not stored in Google Cloud, with keys being transmitted for each operation. Cloud External Key Manager enables clients to store and manage encryption keys in a third-party external key management system outside of Google Cloud infrastructure while still using Google Cloud services for encryption operations.

Why the other options are wrong

  • B. Google default encryption stores keys within Google Cloud, not meeting the requirement that keys be stored outside the CSP.
  • C. Secret Manager stores secrets within Google Cloud, failing the requirement to keep keys outside the cloud service provider.
  • E. Customer-managed encryption keys are still stored within Google Cloud (in Cloud KMS), not meeting the external storage requirement.

Question 12

You have noticed an increased number of phishing attacks across your enterprise user accounts. You want to implement the Google 2-Step Verification (2SV) option that uses a cryptographic signature to authenticate a user and verify the URL of the login page.

Which Google 2SV option should you use?

  1. Titan Security Keys
  2. Google prompt
  3. Google Authenticator app
  4. Cloud HSM keys
Show answer and explanation

Correct answer: A. Titan Security Keys

Titan Security Keys use cryptographic signatures to authenticate users and verify the login page URL, protecting against phishing attacks by ensuring the login session cannot be hijacked even if the user visits a fake phishing site. The key cryptographically verifies the authentic Google login domain, preventing credential theft through phishing.

Why the other options are wrong

  • B. Google Prompt sends notifications to the user's trusted device but does not use cryptographic signatures to verify the login URL.
  • C. Google Authenticator generates time-based one-time passwords but does not verify the login page URL or use cryptographic signatures.
  • D. Cloud HSM keys are for managing encryption keys in the cloud, not for user authentication or 2-step verification.

Question 13

You are designing a new governance model for your organization's secrets that are stored in Secret Manager. Currently, secrets for Production and Non-Production applications are stored and accessed using service accounts. Your proposed solution must:

• Provide granular access to secrets

• Give you control over the rotation schedules for the encryption keys that wrap your secrets

• Maintain environment separation

• Provide ease of management

Which approach should you take?

  1. 1. Use separate Google Cloud projects to store Production and Non-Production secrets. 2. Enforce access control to secrets using projectlevel identity and Access Management (IAM) bindings. 3. Use customer-managed encryption keys to encrypt secrets.
  2. 1. Use a single Google Cloud project to store both Production and Non-Production secrets. 2. Enforce access control to secrets using secret-level Identity and Access Management (IAM) bindings. 3. Use Google-managed encryption keys to encrypt secrets.
  3. 1. Use separate Google Cloud projects to store Production and Non-Production secrets. 2. Enforce access control to secrets using secretlevel Identity and Access Management (IAM) bindings. 3. Use Google-managed encryption keys to encrypt secrets.
  4. 1. Use a single Google Cloud project to store both Production and Non-Production secrets. 2. Enforce access control to secrets using project-level Identity and Access Management (IAM) bindings. 3. Use customer-managed encryption keys to encrypt secrets.
Show answer and explanation

Correct answer: A. 1. Use separate Google Cloud projects to store Production and Non-Production secrets. 2. Enforce access control to secrets using projectlevel identity and Access Management (IAM) bindings. 3. Use customer-managed encryption keys to encrypt secrets.

This approach meets all requirements by using separate projects for Production and Non-Production environments to maintain environment separation and ease of management through project-level organization. Project-level IAM bindings provide granular access control at the environment level while secret-level granularity can be added as needed. Customer-managed encryption keys (CMEK) give you full control over rotation schedules and key residency. This separation prevents accidental cross-environment access and simplifies compliance management.

Why the other options are wrong

  • B. Single project for both environments breaks the separation requirement and makes compliance harder; Google-managed keys prevent control over rotation and residency.
  • C. Google-managed keys do not provide control over rotation schedules or key residency, failing the key control requirement.
  • D. Single project violates environment separation; project-level IAM is less granular than needed for individual secret access control.

Question 14

How should a customer reliably deliver Stackdriver logs from GCP to their on-premises SIEM system?

  1. Send all logs to the SIEM system via an existing protocol such as syslog.
  2. Configure every project to export all their logs to a common BigQuery DataSet, which will be queried by the SIEM system.
  3. Configure Organizational Log Sinks to export logs to a Cloud Pub/Sub Topic, which will be sent to the SIEM via Dataflow.
  4. Build a connector for the SIEM to query for all logs in real time from the GCP RESTful JSON APIs.
Show answer and explanation

Correct answer: C. Configure Organizational Log Sinks to export logs to a Cloud Pub/Sub Topic, which will be sent to the SIEM via Dataflow.

Organizational Log Sinks provide centralized log routing at the organization level. By exporting logs to a Cloud Pub/Sub Topic, organizations can reliably deliver logs to on-premises SIEM systems through Dataflow, which acts as a reliable bridge. This approach handles log buffering, retry logic, and delivery guarantees that are essential for enterprise SIEM integration without requiring individual project configuration.

Why the other options are wrong

  • A. Direct syslog delivery lacks the reliability, buffering, and scalability guarantees needed for production SIEM integration.
  • B. Exporting all logs to BigQuery and having the SIEM query in real time creates latency and puts query load on the SIEM system.
  • D. Real-time RESTful API queries from the SIEM create excessive load on GCP APIs and lack the buffering and reliability guarantees of a message-based approach.

Question 15

An organization wants to track how bonus compensations have changed over time to identify employee outliers and correct earning disparities.

This task must be performed without exposing the sensitive compensation data for any individual and must be reversible to identify the outlier.

Which Cloud Data Loss Prevention API technique should you use?

  1. Cryptographic hashing
  2. Redaction
  3. Format-preserving encryption
  4. Generalization
Show answer and explanation

Correct answer: C. Format-preserving encryption

Format-preserving encryption (FPE) is the correct technique because it encrypts sensitive compensation data while maintaining its format and structure, allowing statistical analysis and trend identification without exposing actual values. The encryption is reversible, enabling identification of outliers by decrypting specific records when needed. Cryptographic hashing is irreversible, making outlier identification impossible. Redaction permanently removes data, preventing any analysis. Generalization reduces precision too much for accurate outlier detection.

Why the other options are wrong

  • A. Cryptographic hashing is irreversible and would prevent the ability to identify specific outliers once hashed.
  • B. Redaction permanently removes sensitive data, making it impossible to perform any subsequent analysis or outlier identification.
  • D. Generalization reduces data precision too significantly to accurately track changes and identify individual compensation outliers.

Question 16

A customer needs to prevent attackers from hijacking their domain/IP and redirecting users to a malicious site through a man-in-the-middle attack.

Which solution should this customer use?

  1. VPC Flow Logs
  2. Cloud Armor
  3. DNS Security Extensions
  4. Cloud Identity-Aware Proxy
Show answer and explanation

Correct answer: C. DNS Security Extensions

DNS Security Extensions (DNSSEC) cryptographically sign DNS records to prevent DNS hijacking and man-in-the-middle attacks where attackers redirect domain/IP traffic to malicious sites. DNSSEC ensures that DNS responses are authentic and have not been tampered with. VPC Flow Logs only monitor network traffic, Cloud Armor protects against DDoS and application attacks, and Cloud Identity-Aware Proxy controls access authentication, none address DNS hijacking.

Why the other options are wrong

  • A. VPC Flow Logs monitor network traffic flow but do not prevent DNS hijacking or domain redirection attacks.
  • B. Cloud Armor protects against DDoS and web application attacks but does not address DNS hijacking or domain/IP redirection.
  • D. Cloud Identity-Aware Proxy controls user authentication and access to applications but does not prevent DNS hijacking.

Question 17

You discovered that sensitive personally identifiable information (PII) is being ingested to your Google Cloud environment in the daily ETL process from an on-premises environment to your BigQuery datasets. You need to redact this data to obfuscate the PII, but need to re-identify it for data analytics purposes.

Which components should you use in your solution? (Choose two.)

  1. Secret Manager
  2. Cloud Key Management Service
  3. Cloud Data Loss Prevention with cryptographic hashing
  4. Cloud Data Loss Prevention with automatic text redaction
  5. Cloud Data Loss Prevention with deterministic encryption using AES-SIV
Show answer and explanation

Correct answer: B, E. Cloud Key Management Service; Cloud Data Loss Prevention with deterministic encryption using AES-SIV

Re-identification requires a reversible transformation. Cloud DLP deterministic encryption with AES-SIV replaces PII with a consistent token that can be decrypted later, and the encryption key is protected by wrapping it with Cloud KMS.

Why the other options are wrong

  • A. Secret Manager stores secrets but is not the key wrapping service Cloud DLP uses for its cryptographic transformations.
  • C. Cryptographic hashing is one-way, so hashed values cannot be re-identified.
  • D. Redaction removes the data permanently, so it cannot be re-identified.

Question 18

You are part of a security team that wants to ensure that a Cloud Storage bucket in Project A can only be readable from Project B. You also want to ensure that data in the Cloud Storage bucket cannot be accessed from or copied to Cloud Storage buckets outside the network, even if the user has the correct credentials.

What should you do?

  1. Enable VPC Service Controls, create a perimeter with Project A and B, and include Cloud Storage service.
  2. Enable Domain Restricted Sharing Organization Policy and Bucket Policy Only on the Cloud Storage bucket.
  3. Enable Private Access in Project A and B networks with strict firewall rules to allow communication between the networks.
  4. Enable VPC Peering between Project A and B networks with strict firewall rules to allow communication between the networks.
Show answer and explanation

Correct answer: A. Enable VPC Service Controls, create a perimeter with Project A and B, and include Cloud Storage service.

VPC Service Controls creates a security perimeter that restricts data access to only resources and identities within the perimeter. By creating a perimeter that includes both Project A and Project B with Cloud Storage service, the bucket becomes accessible only from within that perimeter, preventing data exfiltration to external Cloud Storage buckets or unauthorized projects regardless of user credentials. This prevents both reading and copying data outside the defined security boundary.

Why the other options are wrong

  • B. Domain Restricted Sharing and Bucket Policy Only control IAM permissions and sharing, not network-level data exfiltration prevention.
  • C. Private Access enables private IP connectivity but does not prevent authenticated users with correct credentials from accessing resources outside the network.
  • D. VPC Peering allows network connectivity between projects but does not prevent authenticated API access to resources outside the peered networks.

Question 19

Last week, a company deployed a new App Engine application that writes logs to BigQuery. No other workloads are running in the project. You need to validate that all data written to BigQuery was done using the App Engine Default Service Account.

What should you do?

  1. 1. Use Cloud Logging and filter on BigQuery Insert Jobs. 2. Click on the email address in line with the App Engine Default Service Account in the authentication field. 3. Click Hide Matching Entries. 4. Make sure the resulting list is empty.
  2. 1. Use Cloud Logging and filter on BigQuery Insert Jobs. 2. Click on the email address in line with the App Engine Default Service Account in the authentication field. 3. Click Show Matching Entries. 4. Make sure the resulting list is empty.
  3. 1. In BigQuery, select the related dataset. 2. Make sure that the App Engine Default Service Account is the only account that can write to the dataset.
  4. 1. Go to the Identity and Access Management (IAM) section of the project. 2. Validate that the App Engine Default Service Account is the only account that has a role that can write to BigQuery.
Show answer and explanation

Correct answer: A. 1. Use Cloud Logging and filter on BigQuery Insert Jobs. 2. Click on the email address in line with the App Engine Default Service Account in the authentication field. 3. Click Hide Matching Entries. 4. Make sure the resulting list is empty.

To validate that all BigQuery writes came from the App Engine Default Service Account, use Cloud Logging to filter on BigQuery Insert Jobs, then filter to show only entries from the App Engine Default Service Account by clicking its email. Then use Hide Matching Entries to remove those legitimate entries. If the resulting list is empty, all writes came from the expected service account. Any remaining entries would indicate unauthorized access.

Why the other options are wrong

  • B. Using Show Matching Entries would only display the legitimate entries and wouldn't validate whether other accounts were also writing.
  • C. Checking IAM permissions on a dataset doesn't audit actual data write operations that have already occurred.
  • D. IAM role validation confirms permissions but doesn't audit the actual operations that were performed or verify which accounts actually wrote data.

Question 20

You need to use Cloud External Key Manager to create an encryption key to encrypt specific BigQuery data at rest in Google Cloud.

Which steps should you do first?

  1. 1. Create or use an existing key with a unique uniform resource identifier (URI) in your Google Cloud project. 2. Grant your Google Cloud project access to a supported external key management partner system.
  2. 1. Create or use an existing key with a unique uniform resource identifier (URI) in Cloud Key Management Service (Cloud KMS). 2. In Cloud KMS, grant your Google Cloud project access to use the key.
  3. 1. Create or use an existing key with a unique uniform resource identifier (URI) in a supported external key management partner system. 2. In the external key management partner system, grant access for this key to use your Google Cloud project.
  4. 1. Create an external key with a unique uniform resource identifier (URI) in Cloud Key Management Service (Cloud KMS). 2. In Cloud KMS, grant your Google Cloud project access to use the key.
Show answer and explanation

Correct answer: C. 1. Create or use an existing key with a unique uniform resource identifier (URI) in a supported external key management partner system. 2. In the external key management partner system, grant access for this key to use your Google Cloud project.

When using Cloud External Key Manager (EKM), you must first create or use an existing encryption key in a supported external key management partner system (not in Google Cloud), and then grant that external system access to your Google Cloud project. This architecture keeps the keys outside of Google Cloud infrastructure while allowing Google Cloud services to use them for encryption operations.

Why the other options are wrong

  • A. This reverses the proper sequence; the key must exist in the external partner system first, not in your Google Cloud project.
  • B. Cloud KMS is Google's managed key service and is not used with Cloud External Key Manager, which specifically integrates third-party external key management systems.
  • D. Cloud KMS cannot be used to create external keys; EKM requires keys to be created and managed in the external partner system.

Question 21

A website design company recently migrated all customer sites to App Engine. Some sites are still in progress and should only be visible to customers and company employees from any location.

Which solution will restrict access to the in-progress sites?

  1. Upload an .htaccess file containing the customer and employee user accounts to App Engine.
  2. Create an App Engine firewall rule that allows access from the customer and employee networks and denies all other traffic.
  3. Enable Cloud Identity-Aware Proxy (IAP), and allow access to a Google Group that contains the customer and employee user accounts.
  4. Use Cloud VPN to create a VPN connection between the relevant on-premises networks and the company's GCP Virtual Private Cloud (VPC) network.
Show answer and explanation

Correct answer: C. Enable Cloud Identity-Aware Proxy (IAP), and allow access to a Google Group that contains the customer and employee user accounts.

Cloud Identity-Aware Proxy (IAP) provides application-level access control that authenticates users and enforces authorization based on Google identities. By enabling IAP and restricting access to a Google Group containing customer and employee accounts, the company can control access to in-progress sites from any location without requiring network-level restrictions or VPN connections. This solution integrates seamlessly with Cloud Identity and App Engine.

Why the other options are wrong

  • A. .htaccess files are web server configuration files that don't work with App Engine's application runtime and provide no actual authentication mechanism.
  • B. App Engine firewall rules operate at the network level and cannot authenticate individual users or validate user accounts; they only filter by IP address.
  • D. Cloud VPN restricts access to specific networks but doesn't provide user-level authentication and isn't necessary for controlling access from "any location" as required.

Question 22

A customer wants to move their sensitive workloads to a Compute Engine-based cluster using Managed Instance Groups (MIGs). The jobs are bursty and must be completed quickly. They have a requirement to be able to control the key lifecycle.

Which boot disk encryption solution should you use on the cluster to meet this customer's requirements?

  1. Customer-supplied encryption keys (CSEK)
  2. Customer-managed encryption keys (CMEK) using Cloud Key Management Service (KMS)
  3. Encryption by default
  4. Pre-encrypting files before transferring to Google Cloud Platform (GCP) for analysis
Show answer and explanation

Correct answer: B. Customer-managed encryption keys (CMEK) using Cloud Key Management Service (KMS)

Customer-managed encryption keys (CMEK) using Cloud KMS is the correct solution because it allows customers to control the complete lifecycle of encryption keys while still benefiting from Google Cloud's managed infrastructure. This meets the requirement for key lifecycle control in a Managed Instance Group environment. CSEK requires managing keys outside of Google's systems and is less suitable for managed infrastructure, encryption by default provides no key control, and pre-encryption is inefficient for cloud-based workloads.

Why the other options are wrong

  • A. CSEK requires managing keys externally and is cumbersome for managed instance groups requiring scaling.
  • C. Encryption by default does not provide customer control over the key lifecycle.
  • D. Pre-encrypting files before transfer is operationally inefficient and doesn't leverage cloud-native encryption.

Question 23

A company is running their webshop on Google Kubernetes Engine and wants to analyze customer transactions in BigQuery. You need to ensure that no credit card numbers are stored in BigQuery

What should you do?

  1. Create a BigQuery view with regular expressions matching credit card numbers to query and delete affected rows.
  2. Use the Cloud Data Loss Prevention API to redact related infoTypes before data is ingested into BigQuery.
  3. Leverage Security Command Center to scan for the assets of type Credit Card Number in BigQuery.
  4. Enable Cloud Identity-Aware Proxy to filter out credit card numbers before storing the logs in BigQuery.
Show answer and explanation

Correct answer: B. Use the Cloud Data Loss Prevention API to redact related infoTypes before data is ingested into BigQuery.

The Cloud Data Loss Prevention API is purpose-built to identify and redact sensitive data types, including credit card numbers, before data reaches its destination. By applying DLP policies during the ingestion process into BigQuery, the company ensures credit card data is identified and redacted automatically, preventing sensitive information from ever being stored in the data warehouse.

Why the other options are wrong

  • A. Creating a view and manually querying for credit card patterns is reactive rather than preventive and still allows the data to exist in BigQuery initially.
  • C. Security Command Center scans for security misconfigurations and threats but is not designed to redact or prevent specific data types from being ingested.
  • D. Cloud Identity-Aware Proxy controls access to applications and services; it does not filter data content before storage in BigQuery.

Question 24

An organization is starting to move its infrastructure from its on-premises environment to Google Cloud Platform (GCP). The first step the organization wants to take is to migrate its current data backup and disaster recovery solutions to GCP for later analysis. The organization's production environment will remain on-premises for an indefinite time. The organization wants a scalable and cost-efficient solution.

Which GCP solution should the organization use?

  1. BigQuery using a data pipeline job with continuous updates
  2. Cloud Storage using a scheduled task and gsutil
  3. Compute Engine Virtual Machines using Persistent Disk
  4. Cloud Datastore using regularly scheduled batch upload jobs
Show answer and explanation

Correct answer: B. Cloud Storage using a scheduled task and gsutil

Cloud Storage with scheduled gsutil tasks is the most scalable and cost-efficient solution for migrating backup and disaster recovery data from on-premises to GCP. Cloud Storage is designed for data archival, is highly scalable, offers multiple storage classes for cost optimization, and can be easily automated with scheduled jobs. This approach supports the organization's need for a temporary backup destination while keeping production on-premises.

Why the other options are wrong

  • A. BigQuery is optimized for analytical queries, not for backup storage or disaster recovery solutions.
  • C. Compute Engine with Persistent Disk is expensive and unnecessary for a backup/DR solution that doesn't require compute capacity.
  • D. Cloud Datastore is a NoSQL database; it's not designed for large-scale backup and disaster recovery data storage.

Question 25

An organization adopts Google Cloud Platform (GCP) for application hosting services and needs guidance on setting up password requirements for their Cloud Identity account. The organization has a password policy requirement that corporate employee passwords must have a minimum number of characters.

Which Cloud Identity password guidelines can the organization use to inform their new requirements?

  1. Set the minimum length for passwords to be 8 characters.
  2. Set the minimum length for passwords to be 10 characters.
  3. Set the minimum length for passwords to be 12 characters.
  4. Set the minimum length for passwords to be 6 characters.
Show answer and explanation

Correct answer: A. Set the minimum length for passwords to be 8 characters.

Google Cloud Identity's password guidelines recommend a minimum of 8 characters as the standard requirement for corporate passwords. This aligns with Google-recommended practices for Cloud Identity account security while being practical for enterprise deployment. The 8-character minimum provides adequate security without being overly restrictive.

Why the other options are wrong

  • B. 10 characters exceeds Google Cloud Identity's standard password minimum length recommendation.
  • C. 12 characters exceeds Google Cloud Identity's standard password minimum length recommendation.
  • D. 6 characters is below Google Cloud Identity's recommended minimum length for corporate password security.

25 questions in. Want the rest? The first half of the pack, free by email · The full pack

Question 26

A company allows every employee to use Google Cloud Platform. Each department has a Google Group, with all department members as group members. If a department member creates a new project, all members of that department should automatically have read-only access to all new project resources. Members of any other department should not have access to the project. You need to configure this behavior.

What should you do to meet these requirements?

  1. Create a Folder per department under the Organization. For each department's Folder, assign the Project Viewer role to the Google Group related to that department.
  2. Create a Folder per department under the Organization. For each department's Folder, assign the Project Browser role to the Google Group related to that department.
  3. Create a Project per department under the Organization. For each department's Project, assign the Project Viewer role to the Google Group related to that department.
  4. Create a Project per department under the Organization. For each department's Project, assign the Project Browser role to the Google Group related to that department.
Show answer and explanation

Correct answer: A. Create a Folder per department under the Organization. For each department's Folder, assign the Project Viewer role to the Google Group related to that department.

Creating a Folder per department and assigning the Project Viewer role to each department's Google Group ensures that when any department member creates a new project within their folder, the folder-level IAM binding automatically grants read-only access to all department members. The Folder resource hierarchy in GCP automatically cascades permissions to child projects, and Project Viewer provides read-only access. This approach scales automatically as new projects are created within each folder.

Why the other options are wrong

  • B. Project Browser role does not provide resource-level read access; Project Viewer is needed for resource visibility.
  • C. Creating one Project per department does not allow automatic permissions on new projects created by department members.
  • D. Creating per-department Projects combined with Project Browser role fails to provide proper resource access control.

Question 27

A customer deploys an application to App Engine and needs to check for Open Web Application Security Project (OWASP) vulnerabilities.

Which service should be used to accomplish this?

  1. Cloud Armor
  2. Google Cloud Audit Logs
  3. Web Security Scanner
  4. Anomaly Detection
Show answer and explanation

Correct answer: C. Web Security Scanner

Google Cloud Web Security Scanner is the service specifically designed to identify OWASP Top 10 vulnerabilities and other web application security issues. It automatically scans App Engine applications and other web applications for vulnerabilities including XSS, CSRF, authentication flaws, and other common web application weaknesses. Cloud Armor handles DDoS protection, Audit Logs track access and changes but don't test for vulnerabilities, and Anomaly Detection identifies unusual patterns rather than security flaws.

Why the other options are wrong

  • A. Cloud Armor is designed for DDoS protection and bot management, not vulnerability scanning.
  • B. Audit Logs track who accessed what resources but do not perform security vulnerability assessments.
  • D. Anomaly Detection identifies unusual behavior patterns but does not scan for known OWASP vulnerabilities.

Question 28

Your organization's Google Cloud VMs are deployed via an instance template that configures them with a public IP address in order to host web services for external users. The VMs reside in a service project that is attached to a host (VPC) project containing one custom Shared VPC for the VMs. You have been asked to reduce the exposure of the VMs to the internet while continuing to service external users. You have already recreated the instance template without a public IP address configuration to launch the managed instance group (MIG).

What should you do?

  1. Deploy a Cloud NAT Gateway in the service project for the MIG.
  2. Deploy a Cloud NAT Gateway in the host (VPC) project for the MIG.
  3. Deploy an external HTTP(S) load balancer in the service project with the MIG as a backend.
  4. Deploy an external HTTP(S) load balancer in the host (VPC) project with the MIG as a backend.
Show answer and explanation

Correct answer: C. Deploy an external HTTP(S) load balancer in the service project with the MIG as a backend.

An external HTTP(S) load balancer deployed in the service project with the MIG as a backend provides external users access to the web services while the VMs themselves have no public IPs. The load balancer's public IP handles external traffic, and internal communication to the MIG backends does not require public IPs, reducing internet exposure.

Why the other options are wrong

  • A. Cloud NAT in the service project only handles outbound traffic and does not provide a mechanism for external users to reach inbound web services.
  • B. Cloud NAT in the host project has the same limitation, it addresses outbound traffic, not inbound access for external users to the web services.
  • D. While a load balancer in the host project could work architecturally, the correct practice is to deploy application-facing resources like load balancers in the service project that owns the MIG.

Question 29

An engineering team is launching a web application that will be public on the internet. The web application is hosted in multiple GCP regions and will be directed to the respective backend based on the URL request.

Your team wants to avoid exposing the application directly on the internet and wants to deny traffic from a specific list of malicious IP addresses.

Which solution should your team implement to meet these requirements?

  1. Cloud Armor
  2. Network Load Balancing
  3. SSL Proxy Load Balancing
  4. NAT Gateway
Show answer and explanation

Correct answer: A. Cloud Armor

Cloud Armor is Google Cloud's distributed denial-of-service (DDoS) protection and Web Application Firewall (WAF) service. It allows organizations to create security policies that block traffic from specific IP addresses, protect applications from DDoS attacks, and enforce rate limiting. Cloud Armor works with HTTP(S) load balancers to protect applications while keeping them accessible from the internet without direct exposure.

Why the other options are wrong

  • B. Network Load Balancing distributes traffic but doesn't provide DDoS protection or IP-based filtering capabilities.
  • C. SSL Proxy Load Balancing handles SSL termination and load distribution but lacks DDoS protection and malicious IP blocking features.
  • D. NAT Gateway translates IP addresses for outbound traffic from private networks; it doesn't protect inbound traffic or block malicious IPs.

Question 30

You have stored company approved compute images in a single Google Cloud project that is used as an image repository. This project is protected with VPC Service Controls and exists in the perimeter along with other projects in your organization. This lets other projects deploy images from the image repository project. A team requires deploying a third-party disk image that is stored in an external Google Cloud organization. You need to grant read access to the disk image so that it can be deployed into the perimeter.

What should you do?

  1. Allow the external project by using the organizational policy, constraints/compute.trustedImageProjects.
  2. 1. Update the perimeter. 2. Configure the egressTo field to include the external Google Cloud project number as an allowed resource and the serviceName to compute.googleapis.com. 3. Configure the egressFrom field to set identityType to ANY_IDENTITY.
  3. 1. Update the perimeter. 2. Configure the ingressFrom field to set identityType to ANY_IDENTITY. 3. Configure the ingressTo field to include the external Google Cloud project number as an allowed resource and the serviceName to compute.googleapis.com.
  4. 1. Update the perimeter. 2. Configure the egressTo field to set identityType to ANY_IDENTITY. 3. Configure the egressFrom field to include the external Google Cloud project number as an allowed resource and the serviceName to compute.googleapis.com.
Show answer and explanation

Correct answer: B. 1. Update the perimeter. 2. Configure the egressTo field to include the external Google Cloud project number as an allowed resource and the serviceName to compute.googleapis.com. 3. Configure the egressFrom field to set identityType to ANY_IDENTITY.

When a perimeter needs to access external resources outside the service control perimeter, the egressTo field must be configured to allow traffic to the external project. The egressTo field specifies where traffic is allowed to go, including the external Google Cloud project number and the compute.googleapis.com service. The egressFrom field with ANY_IDENTITY allows any identity within the perimeter to initiate the egress traffic. This configuration permits the internal perimeter projects to read disk images from the external organization's project.

Why the other options are wrong

  • A. Organizational policies for trusted image projects do not work across VPC Service Control perimeter boundaries.
  • C. Ingress rules control traffic coming into the perimeter, not outgoing traffic to external resources.
  • D. The egressFrom and egressTo fields are reversed in this option; egressTo should contain the external project, not egressFrom.

Question 31

Your organization operates Virtual Machines (VMs) with only private IPs in the Virtual Private Cloud (VPC) with internet access through Cloud NAT. Everyday, you must patch all VMs with critical OS updates and provide summary reports.

What should you do?

  1. Validate that the egress firewall rules allow any outgoing traffic. Log in to each VM and execute OS specific update commands. Configure the Cloud Scheduler job to update with critical patches daily for daily updates.
  2. Copy the latest patches to the Cloud Storage bucket. Log in to each VM, download the patches from the bucket, and install them.
  3. Assign public IPs to VMs. Validate that the egress firewall rules allow any outgoing traffic. Log in to each VM, and configure a daily cron job to enable for OS updates at night during low activity periods.
  4. Ensure that VM Manager is installed and running on the VMs. In the OS patch management service, configure the patch jobs to update with critical patches dally.
Show answer and explanation

Correct answer: D. Ensure that VM Manager is installed and running on the VMs. In the OS patch management service, configure the patch jobs to update with critical patches dally.

OS Config agent with VM Manager patch jobs is the correct solution for managing critical patches across private VMs at scale with automated reporting. OS Config provides centralized patch management through the VM Manager console, enabling daily patch jobs with summary reports without requiring manual login to each VM or public IP assignment. This maintains security posture while providing automated compliance. Option A requires manual intervention. Option B is inefficient for large deployments. Option C exposes VMs to the internet, violating the private IP requirement.

Why the other options are wrong

  • A. Manual login and execution on each VM is not scalable and doesn't provide the automated reporting required for daily patching across many VMs.
  • B. Manually downloading patches from Cloud Storage to each VM is inefficient and doesn't provide automated scheduling or summary reporting capabilities.
  • C. Assigning public IPs to VMs contradicts the requirement for private-only infrastructure and increases security risk unnecessarily.

Question 32

For compliance reporting purposes, the internal audit department needs you to provide the list of virtual machines (VMs) that have critical operating system (OS) security updates available, but not installed. You must provide this list every six months, and you want to perform this task quickly.

What should you do?

  1. Run a Security Command Center security scan on all VMs to extract a list of VMs with critical OS vulnerabilities every six months.
  2. Run a gcloud CLI command from the Command Line Interface (CLI) to extract the VM's OS version information every six months.
  3. Ensure that the Cloud Logging agent is installed on all VMs, and extract the OS last update log date every six months.
  4. Ensure the OS Config agent is installed on all VMs and extract the patch status dashboard every six months.
Show answer and explanation

Correct answer: D. Ensure the OS Config agent is installed on all VMs and extract the patch status dashboard every six months.

The OS Config agent with patch status dashboard provides the quickest method to identify VMs with available critical OS security updates. OS Config automatically scans VMs for available patches and displays this information in a dashboard that can be queried every six months for compliance reporting. This is faster and more reliable than other methods because it's purpose-built for patch management. Security Command Center scans vulnerabilities but not specifically update availability. Cloud Logging requires manual log parsing. gcloud CLI commands don't provide comprehensive patch status data.

Why the other options are wrong

  • A. Security Command Center identifies vulnerabilities rather than specifically tracking available OS security updates that haven't been installed.
  • B. gcloud CLI requires manual queries and scripting to extract version information, and doesn't provide direct patch availability status.
  • C. Cloud Logging requires manual extraction and parsing of log dates, which is slower than using the purpose-built OS Config patch status dashboard.

Question 33

An office manager at your small startup company is responsible for matching payments to invoices and creating billing alerts. For compliance reasons, the office manager is only permitted to have the Identity and Access Management (IAM) permissions necessary for these tasks.

Which two IAM roles should the office manager have? (Choose two.)

  1. Organization Administrator
  2. Project Creator
  3. Billing Account Viewer
  4. Billing Account Costs Manager
  5. Billing Account User
Show answer and explanation

Correct answer: C, D. Billing Account Viewer; Billing Account Costs Manager

The office manager needs Billing Account Viewer to view invoices and payment information for matching purposes, and Billing Account Costs Manager to create billing alerts and manage cost tracking. Billing Account Viewer provides read-only access to see billing data, while Billing Account Costs Manager specifically enables creating and managing billing alerts without broader account permissions. These two roles provide the minimum permissions necessary for the stated tasks.

Why the other options are wrong

  • A. Organization Administrator has far too much broad authority across the entire organization, violating the principle of least privilege.
  • B. Project Creator permission is unrelated to matching payments and invoices or creating billing alerts.
  • E. Billing Account User allows resource spending and project linking but does not provide access to view invoices or create billing alerts.

Question 34

You are the Security Admin in your company. You want to synchronize all security groups that have an email address from your LDAP directory in Cloud IAM.

What should you do?

  1. Configure Google Cloud Directory Sync to sync security groups using LDAP search rules that have "user email address" as the attribute to facilitate one-way sync.
  2. Configure Google Cloud Directory Sync to sync security groups using LDAP search rules that have "user email address" as the attribute to facilitate bidirectional sync.
  3. Use a management tool to sync the subset based on the email address attribute. Create a group in the Google domain. A group created in a Google domain will automatically have an explicit Google Cloud Identity and Access Management (IAM) role.
  4. Use a management tool to sync the subset based on group object class attribute. Create a group in the Google domain. A group created in a Google domain will automatically have an explicit Google Cloud Identity and Access Management (IAM) role.
Show answer and explanation

Correct answer: A. Configure Google Cloud Directory Sync to sync security groups using LDAP search rules that have "user email address" as the attribute to facilitate one-way sync.

Google Cloud Directory Sync is the correct tool for synchronizing LDAP security groups into Cloud IAM. Using LDAP search rules with email address as the attribute enables identification and sync of groups that have email addresses. One-way sync is appropriate for this security-focused use case as it prevents unauthorized changes to IAM groups from the LDAP side.

Why the other options are wrong

  • B. Bidirectional sync introduces unnecessary security risk by allowing changes in Cloud IAM to propagate back to the LDAP directory.
  • C. While using a management tool is possible, LDAP search rules should filter by group object class, not user email address, since the requirement is to sync security groups.
  • D. Using group object class attribute is correct for identifying groups, but LDAP search rules based on object class rather than email address would sync all security groups, not just those with email addresses.

Question 35

You need to implement an encryption-at-rest strategy that protects sensitive data and reduces key management complexity for non-sensitive data.

Your solution has the following requirements:

• Schedule key rotation for sensitive data.

• Control which region the encryption keys for sensitive data are stored in.

• Minimize the latency to access encryption keys for both sensitive and non-sensitive data.

What should you do?

  1. Encrypt non-sensitive data and sensitive data with Cloud External Key Manager.
  2. Encrypt non-sensitive data and sensitive data with Cloud Key Management Service.
  3. Encrypt non-sensitive data with Google default encryption, and encrypt sensitive data with Cloud External Key Manager.
  4. Encrypt non-sensitive data with Google default encryption, and encrypt sensitive data with Cloud Key Management Service.
Show answer and explanation

Correct answer: D. Encrypt non-sensitive data with Google default encryption, and encrypt sensitive data with Cloud Key Management Service.

The optimal approach encrypts non-sensitive data with Google's default encryption (which requires minimal management overhead) and sensitive data with Cloud KMS, which provides scheduled key rotation, regional key storage control, and low-latency access within Google Cloud. This two-tiered strategy reduces key management complexity for non-sensitive data while meeting all requirements for sensitive data protection. Cloud External Key Manager adds unnecessary complexity and increases latency for sensitive data that resides in Google Cloud.

Why the other options are wrong

  • A. Using Cloud External Key Manager for non-sensitive data adds unnecessary key management complexity and latency without corresponding security benefits.
  • B. While Cloud KMS can meet the requirements, using it for non-sensitive data increases complexity unnecessarily when Google default encryption is sufficient.
  • C. Cloud External Key Manager introduces higher latency and management complexity that exceeds the requirements for the sensitive data use case in this scenario.

Question 36

Your organization develops software involved in many open source projects and is concerned about software supply chain threats. You need to deliver provenance for the build to demonstrate the software is untampered.

What should you do?

  1. 1. Hire an external auditor to review and provide provenance. 2. Define the scope and conditions. 3. Get support from the Security department or representative. 4. Publish the attestation to your public web page.
  2. 1. Review the software process. 2. Generate private and public key pairs and use Pretty Good Privacy (PGP) protocols to sign the output software artifacts together with a file containing the address of your enterprise and point of contact. 3. Publish the PGP signed attestation to your public web page.
  3. 1. Publish the software code on GitHub as open source. 2. Establish a bug bounty program, and encourage the open source community to review, report, and fix the vulnerabilities.
  4. 1. Generate Supply Chain Levels for Software Artifacts (SLSA) level 3 assurance by using Cloud Build. 2. View the build provenance in the Security insights side panel within the Google Cloud console.
Show answer and explanation

Correct answer: D. 1. Generate Supply Chain Levels for Software Artifacts (SLSA) level 3 assurance by using Cloud Build. 2. View the build provenance in the Security insights side panel within the Google Cloud console.

Cloud Build with SLSA level 3 assurance is the correct approach for delivering provenance in software supply chain security. SLSA (Supply Chain Levels for Software Artifacts) provides a standardized framework for software artifact integrity, and Cloud Build automatically generates cryptographically signed build provenance that can be viewed in Security insights. This addresses supply chain threats by proving the software is untampered and built through a controlled, auditable process. Option A lacks technical rigor. Option B uses outdated PGP signing without modern provenance standards. Option C relies on community review rather than deterministic provenance.

Why the other options are wrong

  • A. External auditor review and manual attestation processes lack the automated, cryptographic provenance that modern supply chain security requires.
  • B. PGP signing alone does not provide the comprehensive SLSA provenance metadata needed to demonstrate the complete build process was untampered.
  • C. Bug bounty programs address vulnerabilities after publication but do not deliver the build provenance needed to prove supply chain integrity.

Question 37

An organization's typical network and security review consists of analyzing application transit routes, request handling, and firewall rules. They want to enable their developer teams to deploy new applications without the overhead of this full review.

How should you advise this organization?

  1. Use Forseti with Firewall filters to catch any unwanted configurations in production.
  2. Mandate use of infrastructure as code and provide static analysis in the CI/CD pipelines to enforce policies.
  3. Route all VPC traffic through customer-managed routers to detect malicious patterns in production.
  4. All production applications will run on-premises. Allow developers free rein in GCP as their dev and QA platforms.
Show answer and explanation

Correct answer: B. Mandate use of infrastructure as code and provide static analysis in the CI/CD pipelines to enforce policies.

Mandating infrastructure as code with static analysis in CI/CD pipelines enables developers to deploy applications safely without blocking on manual review processes. Policy enforcement happens automatically during the build pipeline, catching network, firewall, and security misconfigurations before they reach production. This shifts compliance left and accelerates deployment while maintaining security standards. Forseti is post-deployment detection, custom routers add operational complexity, and allowing unbounded dev environments defeats the purpose of governance.

Why the other options are wrong

  • A. Forseti detects issues in production after deployment rather than preventing them proactively in the pipeline.
  • C. Customer-managed routers for all traffic inspection adds significant operational overhead without automating policy compliance.
  • D. Allowing free rein in GCP dev environments defeats security governance and creates a false divide between dev and production security practices.

Question 38

An organization receives an increasing number of phishing emails.

Which method should be used to protect employee credentials in this situation?

  1. Multifactor Authentication
  2. A strict password policy
  3. Captcha on login pages
  4. Encrypted emails
Show answer and explanation

Correct answer: A. Multifactor Authentication

Multifactor Authentication (MFA) is the most effective defense against phishing attacks. Even if attackers obtain credentials through phishing, they cannot access accounts without the second authentication factor. MFA prevents unauthorized access regardless of password compromise, making it the primary technical control for protecting against credential theft from phishing.

Why the other options are wrong

  • B. Strict password policies alone do not prevent users from willingly providing credentials to phishers or protect against password reuse across services.
  • C. Captcha prevents automated attacks but does not stop users from voluntarily entering credentials on phishing pages.
  • D. Encrypted emails do not protect credentials that users may voluntarily share with attackers through phishing interactions.

Question 39

Your team uses a service account to authenticate data transfers from a given Compute Engine virtual machine instance of to a specified Cloud Storage bucket. An engineer accidentally deletes the service account, which breaks application functionality. You want to recover the application as quickly as possible without compromising security.

What should you do?

  1. Temporarily disable authentication on the Cloud Storage bucket.
  2. Use the undelete command to recover the deleted service account.
  3. Create a new service account with the same name as the deleted service account.
  4. Update the permissions of another existing service account and supply those credentials to the applications.
Show answer and explanation

Correct answer: B. Use the undelete command to recover the deleted service account.

GCP provides an undelete command that recovers recently deleted service accounts within a grace period, restoring the account with all its original credentials and permissions. This is the fastest and most secure recovery method that preserves the original authentication setup.

Why the other options are wrong

  • A. Disabling authentication on the Cloud Storage bucket exposes data to unauthorized access and violates security requirements.
  • C. Creating a new service account with the same name does not recover the deleted account's credentials and keys, requiring complete reconfiguration and credential rotation.
  • D. Using another existing service account requires updating all applications and introduces unnecessary complexity when the original account can be recovered.

Question 40

You work for a large organization where each business unit has thousands of users. You need to delegate management of access control permissions to each business unit. You have the following requirements:

• Each business unit manages access controls for their own projects.

• Each business unit manages access control permissions at scale.

• Business units cannot access other business units' projects.

• Users lose their access if they move to a different business unit or leave the company.

• Users and access control permissions are managed by the on-premises directory service.

What should you do? (Choose two.)

  1. Use VPC Service Controls to create perimeters around each business unit's project.
  2. Organize projects in folders, and assign permissions to Google groups at the folder level.
  3. Group business units based on Organization Units (OUs) and manage permissions based on OUs
  4. Create a project naming convention, and use Google's IAM Conditions to manage access based on the prefix of project names.
  5. Use Google Cloud Directory Sync to synchronize users and group memberships in Cloud Identity.
Show answer and explanation

Correct answer: B, E. Organize projects in folders, and assign permissions to Google groups at the folder level; Use Google Cloud Directory Sync to synchronize users and group memberships in Cloud Identity.

Organizing projects in folders and assigning permissions to Google groups at the folder level (option B) enables each business unit to manage their own projects while preventing cross-unit access and simplifying permission management at scale. Using Google Cloud Directory Sync to synchronize users and group memberships from the on-premises directory (option E) ensures that user access is automatically revoked when users leave or change business units, and keeps user identities in sync with the authoritative on-premises source.

Why the other options are wrong

  • A. VPC Service Controls creates security perimeters but does not directly solve the delegation and group-based access control requirements.
  • C. Managing permissions based on on-premises OUs requires synchronization to Cloud Identity first; OUs alone do not manage cloud permissions.
  • D. Using project naming conventions with IAM Conditions is not a scalable approach for managing permissions across thousands of users and does not leverage the on-premises directory.

Question 41

A customer deployed an application on Compute Engine that takes advantage of the elastic nature of cloud computing.

How can you work with Infrastructure Operations Engineers to best ensure that Windows Compute Engine VMs are up to date with all the latest OS patches?

  1. Build new base images when patches are available, and use a CI/CD pipeline to rebuild VMs, deploying incrementally.
  2. Federate a Domain Controller into Compute Engine, and roll out weekly patches via Group Policy Object.
  3. Use Deployment Manager to provision updated VMs into new serving Instance Groups (IGs).
  4. Reboot all VMs during the weekly maintenance window and allow the StartUp Script to download the latest patches from the internet.
Show answer and explanation

Correct answer: A. Build new base images when patches are available, and use a CI/CD pipeline to rebuild VMs, deploying incrementally.

Building new base images when patches are available and using a CI/CD pipeline to rebuild and incrementally deploy VMs is the best practice for elastic cloud infrastructure. This approach leverages immutable infrastructure principles, ensures consistent patching across all instances, allows testing before production deployment, and aligns with cloud-native operations. It avoids manual patching and ensures instances are always created from known-good images.

Why the other options are wrong

  • B. Federating Active Directory into Compute Engine adds unnecessary complexity and ties cloud infrastructure to on-premises management patterns inappropriate for elastic cloud.
  • C. Deployment Manager creates new instances but does not address patching existing instances or automate the patching process.
  • D. Relying on startup scripts to download patches is unreliable, introduces inconsistency, and does not follow cloud best practices for immutable infrastructure.

Question 42

A company has redundant mail servers in different Google Cloud Platform regions and wants to route customers to the nearest mail server based on location.

How should the company accomplish this?

  1. Configure TCP Proxy Load Balancing as a global load balancing service listening on port 995.
  2. Create a Network Load Balancer to listen on TCP port 995 with a forwarding rule to forward traffic based on location.
  3. Use Cross-Region Load Balancing with an HTTP(S) load balancer to route traffic to the nearest region.
  4. Use Cloud CDN to route the mail traffic to the closest origin mail server based on client IP address.
Show answer and explanation

Correct answer: A. Configure TCP Proxy Load Balancing as a global load balancing service listening on port 995.

TCP Proxy Load Balancing is a global load balancing service that can route traffic based on geographic location to the nearest backend in different regions. It supports TCP traffic on port 995 (IMAPS) and provides automatic routing to the nearest healthy mail server across regions, which is exactly what's needed for location-based routing of mail servers.

Why the other options are wrong

  • B. Network Load Balancer is regional, not global, so it cannot route traffic across multiple regions based on client location.
  • C. HTTP(S) load balancer is designed for HTTP/HTTPS traffic, not mail protocols like IMAPS on port 995.
  • D. Cloud CDN is optimized for caching and serving content; it's not designed for routing mail server traffic based on location in the way required here.

Question 43

Your company is using GSuite and has developed an application meant for internal usage on Google App Engine. You need to make sure that an external user cannot gain access to the application even when an employee's password has been compromised.

What should you do?

  1. Enforce 2-factor authentication in GSuite for all users.
  2. Configure Cloud Identity-Aware Proxy for the App Engine Application.
  3. Provision user passwords using GSuite Password Sync.
  4. Configure Cloud VPN between your private network and GCP.
Show answer and explanation

Correct answer: A. Enforce 2-factor authentication in GSuite for all users.

If a password is compromised, the attacker still needs the second factor to sign in, so enforcing 2-step verification in G Suite stops the external user. Identity-Aware Proxy only checks that the user has authenticated with a valid Google identity, which a stolen password alone could satisfy.

Why the other options are wrong

  • B. IAP relies on the user's Google sign-in, so a stolen password without a second factor could still pass IAP.
  • C. Password Sync only keeps passwords consistent between directories and does nothing to block a stolen password.
  • D. A VPN secures network connectivity but does not address the use of stolen user credentials for an App Engine app.

Question 44

A manager wants to start retaining security event logs for 2 years while minimizing costs. You write a filter to select the appropriate log entries.

Where should you export the logs?

  1. BigQuery datasets
  2. Cloud Storage buckets
  3. StackDriver logging
  4. Cloud Pub/Sub topics
Show answer and explanation

Correct answer: B. Cloud Storage buckets

Cloud Storage buckets are the most cost-effective option for long-term log retention. Cloud Storage supports lifecycle policies to transition data to cheaper storage classes over time, making it ideal for 2-year retention requirements while minimizing costs. BigQuery would be significantly more expensive for long-term log storage, and StackDriver Logging has limited retention periods.

Why the other options are wrong

  • A. BigQuery is optimized for querying and analysis, not for long-term archival storage; it is much more expensive than Cloud Storage for this use case.
  • C. StackDriver Logging (now Cloud Logging) typically retains logs for a limited period; long-term retention requires export to another service like Cloud Storage.
  • D. Cloud Pub/Sub is a messaging service designed for real-time streaming, not for log retention or archival.

Question 45

A customer has 300 engineers. The company wants to grant different levels of access and efficiently manage IAM permissions between users in the development and production environment projects.

Which two steps should the company take to meet these requirements? (Choose two.)

  1. Create a project with multiple VPC networks for each environment.
  2. Create a folder for each development and production environment.
  3. Create a Google Group for the Engineering team, and assign permissions at the folder level.
  4. Create an Organizational Policy constraint for each folder environment.
  5. Create projects for each environment, and grant IAM rights to each engineering user.
Show answer and explanation

Correct answer: B, C. Create a folder for each development and production environment; Create a Google Group for the Engineering team, and assign permissions at the folder level.

Creating a folder for each development and production environment provides the organizational structure needed to manage resources separately. Creating a Google Group for the Engineering team and assigning permissions at the folder level allows efficient management of IAM permissions for all 300 engineers, as folder-level permissions automatically cascade to all projects within that folder. This approach scales efficiently without requiring per-user permission management.

Why the other options are wrong

  • A. Multiple VPC networks within a single project do not provide the isolation and permission management structure needed for different environments.
  • D. Organizational Policy constraints enforce restrictions but do not directly manage IAM permissions or access levels.
  • E. Granting IAM rights to each of 300 individual users is inefficient and violates the requirement to efficiently manage permissions through groups.

Question 46

You are responsible for managing your company's identities in Google Cloud. Your company enforces 2-Step Verification (2SV) for all users. You need to reset a user's access, but the user lost their second factor for 2SV. You want to minimize risk.

What should you do?

  1. On the Google Admin console, select the appropriate user account, and generate a backup code to allow the user to sign in. Ask the user to update their second factor.
  2. On the Google Admin console, temporarily disable the 2SV requirements for all users. Ask the user to log in and add their new second factor to their account. Re-enable the 2SV requirement for all users.
  3. On the Google Admin console, select the appropriate user account, and temporarily disable 2SV for this account. Ask the user to update their second factor, and then re-enable 2SV for this account.
  4. On the Google Admin console, use a super administrator account to reset the user account's credentials. Ask the user to update their credentials after their first login.
Show answer and explanation

Correct answer: A. On the Google Admin console, select the appropriate user account, and generate a backup code to allow the user to sign in. Ask the user to update their second factor.

Generating a backup code for the specific user account minimizes risk by allowing only that user to regain access without disabling 2SV for others or reducing security organization-wide. This targeted approach maintains security posture while solving the individual user's access problem, and the user can then update their second factor with their account still protected.

Why the other options are wrong

  • B. Temporarily disabling 2SV for all users creates a widespread security vulnerability affecting the entire organization unnecessarily.
  • C. Disabling 2SV for the account, even temporarily, leaves that user's account without multi-factor protection during the recovery window.
  • D. Resetting credentials through a super administrator account is more disruptive than necessary and requires the user to change credentials when the actual issue is just the second factor.

Question 47

You plan to use a Google Cloud Armor policy to prevent common attacks such as cross-site scripting (XSS) and SQL injection (SQLi) from reaching your web application's backend.

What are two requirements for using Google Cloud Armor security policies? (Choose two.)

  1. The load balancer must be an external SSL proxy load balancer.
  2. Google Cloud Armor Policy rules can only match on Layer 7 (L7) attributes.
  3. The load balancer must use the Premium Network Service Tier.
  4. The backend service's load balancing scheme must be EXTERNAL.
  5. The load balancer must be an external HTTP(S) load balancer.
Show answer and explanation

Correct answer: D, E. The backend service's load balancing scheme must be EXTERNAL; The load balancer must be an external HTTP(S) load balancer.

Google Cloud Armor policies require an external HTTP(S) load balancer (not SSL proxy) and the backend service must use an EXTERNAL load balancing scheme. These are the foundational requirements for Cloud Armor to inspect and filter traffic at Layer 7. While Cloud Armor can match on Layer 7 attributes, it is not limited to only Layer 7 and operates across multiple layers. Premium Network Service Tier is not a requirement for Cloud Armor functionality.

Why the other options are wrong

  • A. Cloud Armor works with external HTTP(S) load balancers, not SSL proxy load balancers.
  • B. Cloud Armor can match on Layer 7 attributes but is not limited to Layer 7 only.
  • C. Standard Network Service Tier is sufficient for Cloud Armor; Premium tier is not required.

Question 48

A customer wants to make it convenient for their mobile workforce to access a CRM web interface that is hosted on Google Cloud Platform (GCP). The CRM can only be accessed by someone on the corporate network. The customer wants to make it available over the internet. Your team requires an authentication layer in front of the application that supports two-factor authentication

Which GCP product should the customer implement to meet these requirements?

  1. Cloud Identity-Aware Proxy
  2. Cloud Armor
  3. Cloud Endpoints
  4. Cloud VPN
Show answer and explanation

Correct answer: A. Cloud Identity-Aware Proxy

Cloud Identity-Aware Proxy (IAP) provides an authentication and authorization layer in front of applications, controlling access based on identity and context. IAP natively supports two-factor authentication and integrates with identity providers to protect applications from unauthorized access. It allows organizations to make internal applications available over the internet securely without requiring VPN access.

Why the other options are wrong

  • B. Cloud Armor is a DDoS protection service that filters traffic at the network edge, not an authentication layer.
  • C. Cloud Endpoints manages API access and authentication but is not designed for web interface access control with two-factor authentication.
  • D. Cloud VPN creates a tunnel but doesn't provide application-level authentication or two-factor authentication features.

Question 49

Your company runs a website that will store PII on Google Cloud Platform. To comply with data privacy regulations, this data can only be stored for a specific amount of time and must be fully deleted after this specific period. Data that has not yet reached the time period should not be deleted. You want to automate the process of complying with this regulation.

What should you do?

  1. Store the data in a single Persistent Disk, and delete the disk at expiration time.
  2. Store the data in a single BigQuery table and set the appropriate table expiration time.
  3. Store the data in a single Cloud Storage bucket and configure the bucket's Time to Live.
  4. Store the data in a single BigTable table and set an expiration time on the column families.
Show answer and explanation

Correct answer: C. Store the data in a single Cloud Storage bucket and configure the bucket's Time to Live.

A Cloud Storage lifecycle rule with an Age condition acts as a time to live for each object. Every object is deleted when it reaches the configured age, while newer objects that have not reached the retention period are kept. This automates the retention rule without deleting unexpired data.

Why the other options are wrong

  • A. Deleting a whole disk at one time would remove data that has not yet reached its retention period, and it is not automated.
  • B. Table expiration deletes the entire table at once, including rows that have not reached the retention period.
  • D. Bigtable garbage collection works per cell and is not a guaranteed immediate deletion, so it is not the intended simple compliance solution here.

Question 50

Your security team wants to reduce the risk of user-managed keys being mismanaged and compromised. To achieve this, you need to prevent developers from creating user-managed service account keys for projects in their organization.

How should you enforce this?

  1. Configure Secret Manager to manage service account keys.
  2. Enable an organization policy to disable service accounts from being created.
  3. Enable an organization policy to prevent service account keys from being created.
  4. Remove the iam.serviceAccounts.getAccessToken permission from users.
Show answer and explanation

Correct answer: C. Enable an organization policy to prevent service account keys from being created.

An organization policy that prevents service account keys from being created directly addresses the requirement to stop developers from creating user-managed service account keys. This policy-based approach enforces the security control across the organization at scale without requiring individual permission removals or manual management.

Why the other options are wrong

  • A. Secret Manager is used to manage secrets but does not prevent the creation of service account keys at the organizational level.
  • B. Disabling service account creation prevents using service accounts entirely, which is too restrictive; the goal is only to prevent key creation.
  • D. Removing iam.serviceAccounts.getAccessToken permission prevents token access but does not prevent the creation of new keys.

50 questions in. Want the rest? The first half of the pack, free by email · The full pack

Question 51

Your company wants to determine what products they can build to help customers improve their credit scores depending on their age range. To achieve this, you need to join user information in the company's banking app with customers' credit score data received from a third party. While using this raw data will allow you to complete this task, it exposes sensitive data, which could be propagated into new systems.

This risk needs to be addressed using de-identification and tokenization with Cloud Data Loss Prevention while maintaining the referential integrity across the database.

Which cryptographic token format should you use to meet these requirements?

  1. Deterministic encryption
  2. Secure, key-based hashes
  3. Format-preserving encryption
  4. Cryptographic hashing
Show answer and explanation

Correct answer: A. Deterministic encryption

Deterministic encryption is the correct cryptographic token format for this use case because it maintains referential integrity across databases. When the same sensitive value (like a user ID or credit score) is encrypted, deterministic encryption produces the same ciphertext every time, allowing you to join tables and maintain relationships while protecting sensitive data. This enables matching payments to invoices and maintaining data relationships across systems.

Why the other options are wrong

  • B. Secure key-based hashes are non-reversible and cannot maintain the referential integrity needed to join data across databases.
  • C. Format-preserving encryption maintains input format but does not guarantee the deterministic property needed for consistent joins across multiple systems.
  • D. Cryptographic hashing is one-way and irreversible, making it impossible to maintain referential integrity for data joining operations.

Question 52

You are the security admin of your company. You have 3,000 objects in your Cloud Storage bucket. You do not want to manage access to each object individually.

You also do not want the uploader of an object to always have full control of the object. However, you want to use Cloud Audit Logs to manage access to your bucket.

What should you do?

  1. Set up an ACL with OWNER permission to a scope of allUsers.
  2. Set up an ACL with READER permission to a scope of allUsers.
  3. Set up a default bucket ACL and manage access for users using IAM.
  4. Set up Uniform bucket-level access on the Cloud Storage bucket and manage access for users using IAM.
Show answer and explanation

Correct answer: D. Set up Uniform bucket-level access on the Cloud Storage bucket and manage access for users using IAM.

Uniform bucket-level access removes individual object ACLs and manages all access through IAM roles, which eliminates the need to manage 3,000 objects individually and prevents uploaders from having automatic full control. This approach integrates with Cloud Audit Logs through IAM activity logging. Cloud Audit Logs automatically captures IAM-based access decisions, providing the audit trail required for security administration.

Why the other options are wrong

  • A. Setting OWNER permission to allUsers would grant full control to everyone, violating the security requirement to prevent unrestricted access.
  • B. Setting READER permission to allUsers grants read access to all users without authentication, violating the security principle of least privilege.
  • C. Default bucket ACLs still rely on object-level ACL management and don't provide centralized IAM-based access control with full audit logging integration.

Question 53

A retail customer allows users to upload comments and product reviews. The customer needs to make sure the text does not include sensitive data before the comments or reviews are published.

Which Google Cloud Service should be used to achieve this?

  1. Cloud Key Management Service
  2. Cloud Data Loss Prevention API
  3. BigQuery
  4. Web Security Scanner
Show answer and explanation

Correct answer: B. Cloud Data Loss Prevention API

The Cloud Data Loss Prevention API is specifically designed to detect and redact sensitive data patterns such as credit card numbers, social security numbers, and other personally identifiable information. It scans text content to identify sensitive data before publication, making it the ideal service for validating user-generated content like comments and reviews. The other options serve different purposes and are not designed for content scanning.

Why the other options are wrong

  • A. Cloud Key Management Service manages encryption keys rather than detecting sensitive data.
  • C. BigQuery is a data warehouse for analysis and does not provide real-time sensitive data detection.
  • D. Web Security Scanner identifies web application vulnerabilities rather than sensitive data within user content.

Question 54

You need to implement an encryption at-rest strategy that reduces key management complexity for non-sensitive data and protects sensitive data while providing the flexibility of controlling the key residency and rotation schedule. FIPS 140-2 L1 compliance is required for all data types.

What should you do?

  1. Encrypt non-sensitive data and sensitive data with Cloud External Key Manager.
  2. Encrypt non-sensitive data and sensitive data with Cloud Key Management Service
  3. Encrypt non-sensitive data with Google default encryption, and encrypt sensitive data with Cloud External Key Manager.
  4. Encrypt non-sensitive data with Google default encryption, and encrypt sensitive data with Cloud Key Management Service.
Show answer and explanation

Correct answer: D. Encrypt non-sensitive data with Google default encryption, and encrypt sensitive data with Cloud Key Management Service.

For this scenario, non-sensitive data should use Google's default encryption to reduce management complexity, while sensitive data should use Cloud KMS for key control and rotation flexibility. Cloud KMS is FIPS 140-2 L1 compliant and provides customer control over key residency and rotation schedules. Cloud External Key Manager adds unnecessary complexity for non-sensitive data and is designed for scenarios requiring keys to remain outside Google Cloud entirely.

Why the other options are wrong

  • A. Cloud External Key Manager for non-sensitive data adds unnecessary key management complexity.
  • B. Using Cloud KMS for all data, including non-sensitive, increases management overhead without corresponding security benefit.
  • C. Google default encryption is not FIPS 140-2 L1 compliant, violating the stated compliance requirement for all data types.

Question 55

You need to follow Google-recommended practices to leverage envelope encryption and encrypt data at the application layer.

What should you do?

  1. Generate a data encryption key (DEK) locally to encrypt the data, and generate a new key encryption key (KEK) in Cloud KMS to encrypt the DEK. Store both the encrypted data and the encrypted DEK.
  2. Generate a data encryption key (DEK) locally to encrypt the data, and generate a new key encryption key (KEK) in Cloud KMS to encrypt the DEK. Store both the encrypted data and the KEK.
  3. Generate a new data encryption key (DEK) in Cloud KMS to encrypt the data, and generate a key encryption key (KEK) locally to encrypt the key. Store both the encrypted data and the encrypted DEK.
  4. Generate a new data encryption key (DEK) in Cloud KMS to encrypt the data, and generate a key encryption key (KEK) locally to encrypt the key. Store both the encrypted data and the KEK.
Show answer and explanation

Correct answer: A. Generate a data encryption key (DEK) locally to encrypt the data, and generate a new key encryption key (KEK) in Cloud KMS to encrypt the DEK. Store both the encrypted data and the encrypted DEK.

Google-recommended envelope encryption practices require generating the Data Encryption Key (DEK) locally to encrypt application data, then wrapping that DEK with a Key Encryption Key (KEK) created in Cloud KMS. Both the encrypted data and the encrypted DEK are stored together, allowing the application to request decryption of the DEK from Cloud KMS when needed. This separation of concerns protects the actual encryption keys while maintaining cryptographic security.

Why the other options are wrong

  • B. Storing the unencrypted KEK defeats the purpose of envelope encryption by leaving the key encryption key exposed.
  • C. Generating the DEK in Cloud KMS contradicts the application-layer encryption pattern, and having a locally-managed KEK eliminates the security benefits of Cloud KMS.
  • D. Generating the DEK in Cloud KMS is not the recommended pattern for application-layer encryption, and storing an unencrypted KEK locally is insecure.

Question 56

Your team needs to prevent users from creating projects in the organization. Only the DevOps team should be allowed to create projects on behalf of the requester.

Which two tasks should your team perform to handle this request? (Choose two.)

  1. Remove all users from the Project Creator role at the organizational level.
  2. Create an Organization Policy constraint, and apply it at the organizational level.
  3. Grant the Project Editor role at the organizational level to a designated group of users.
  4. Add a designated group of users to the Project Creator role at the organizational level.
  5. Grant the billing account creator role to the designated DevOps team.
Show answer and explanation

Correct answer: A, D. Remove all users from the Project Creator role at the organizational level; Add a designated group of users to the Project Creator role at the organizational level.

To prevent users from creating projects while allowing only DevOps to do so requires two actions. First, remove all users from the Project Creator role at the organizational level to prevent general project creation. Second, grant the Project Creator role to the designated DevOps team group at the organizational level. This combination ensures only authorized users can create projects on behalf of requesters.

Why the other options are wrong

  • B. Organization Policy constraints can restrict project creation but cannot selectively grant creation rights to specific teams; they only deny or restrict.
  • C. The Project Editor role is not the appropriate role for granting project creation permissions.
  • E. The billing account creator role is unrelated to controlling project creation permissions.

Question 57

A customer has an analytics workload running on Compute Engine that should have limited internet access.

Your team created an egress firewall rule to deny (priority 1000) all traffic to the internet.

The Compute Engine instances now need to reach out to the public repository to get security updates.

What should your team do?

  1. Create an egress firewall rule to allow traffic to the CIDR range of the repository with a priority greater than 1000.
  2. Create an egress firewall rule to allow traffic to the CIDR range of the repository with a priority less than 1000.
  3. Create an egress firewall rule to allow traffic to the hostname of the repository with a priority greater than 1000.
  4. Create an egress firewall rule to allow traffic to the hostname of the repository with a priority less than 1000.
Show answer and explanation

Correct answer: B. Create an egress firewall rule to allow traffic to the CIDR range of the repository with a priority less than 1000.

In Google Cloud firewall rules, lower priority numbers are evaluated first. Since the deny rule has priority 1000, you must create an allow rule with a priority less than 1000 (such as 100 or 500) to ensure it is evaluated before the deny rule. This allows traffic to the repository while the higher-numbered deny rule catches all other internet traffic. The answer must specify a CIDR range rather than a hostname because firewall rules operate at the network layer and work with IP addresses, not domain names.

Why the other options are wrong

  • A. Priority greater than 1000 means the deny rule is evaluated first, so the allow rule would never take effect.
  • C. Firewall rules cannot filter by hostname; they only work with CIDR ranges and IP addresses at the network layer.
  • D. Firewall rules cannot filter by hostname; they only work with CIDR ranges and IP addresses at the network layer.

Question 58

A customer is collaborating with another company to build an application on Compute Engine. The customer is building the application tier in their GCP Organization, and the other company is building the storage tier in a different GCP Organization. This is a 3-tier web application. Communication between portions of the application must not traverse the public internet by any means.

Which connectivity option should be implemented?

  1. VPC peering
  2. Cloud VPN
  3. Cloud Interconnect
  4. Shared VPC
Show answer and explanation

Correct answer: A. VPC peering

VPC Network Peering connects VPC networks even when they belong to different projects and different organizations. Traffic between peered networks stays on Google's internal network and never crosses the public internet.

Why the other options are wrong

  • B. Cloud VPN tunnels run over public IP addresses, so the traffic crosses the internet even though it is encrypted.
  • C. Cloud Interconnect is for connecting on-premises networks to Google Cloud, not two Google Cloud organizations.
  • D. Shared VPC works only within a single organization, so it cannot link projects in two different organizations.

Question 59

You are setting up a CI/CD pipeline to deploy containerized applications to your production clusters on Google Kubernetes Engine (GKE). You need to prevent containers with known vulnerabilities from being deployed. You have the following requirements for your solution:

• Must be cloud-native

• Must be cost-efficient

• Minimize operational overhead

How should you accomplish this? (Choose two.)

  1. Create a Cloud Build pipeline that will monitor changes to your container templates in a Cloud Source Repositories repository. Add a step to analyze Container Analysis results before allowing the build to continue.
  2. Use a Cloud Function triggered by log events in Google Cloud's operations suite to automatically scan your container images in Container Registry.
  3. Use a cron job on a Compute Engine instance to scan your existing repositories for known vulnerabilities and raise an alert if a noncompliant container image is found.
  4. Deploy Jenkins on GKE and configure a CI/CD pipeline to deploy your containers to Container Registry. Add a step to validate your container images before deploying your container to the cluster.
  5. In your CI/CD pipeline, add an attestation on your container image when no vulnerabilities have been found. Use a Binary Authorization policy to block deployments of containers with no attestation in your cluster.
Show answer and explanation

Correct answer: A, E. Create a Cloud Build pipeline that will monitor changes to your container templates in a Cloud Source Repositories repository. Add a step to analyze Container Analysis results before allowing the build to continue; In your CI/CD pipeline, add an attestation on your container image when no vulnerabilities have been found. Use a Binary Authorization policy to block deployments of containers with no attestation in your cluster.

Cloud Build integrated with Container Analysis (Option A) provides a cloud-native, cost-efficient CI/CD solution that scans containers before deployment with minimal operational overhead. Binary Authorization (Option E) enforces that only attested, vulnerability-free images can be deployed to GKE clusters, providing an additional security gate. Together, these solutions prevent vulnerable containers from reaching production. Cloud Functions with registry scanning adds operational overhead. A Compute Engine cron job is not cloud-native and requires management. Jenkins on GKE duplicates existing Google Cloud solutions.

Why the other options are wrong

  • B. Using Cloud Functions triggered by log events for scanning adds operational complexity and ongoing management overhead, and is less integrated than native Cloud Build solutions.
  • C. A cron job on Compute Engine is not cloud-native, requires manual management, and is not cost-efficient compared to serverless solutions.
  • D. Deploying Jenkins on GKE duplicates functionality already provided by Cloud Build, adding operational overhead and complexity rather than minimizing it.

Question 60

You perform a security assessment on a customer architecture and discover that multiple VMs have public IP addresses. After providing a recommendation to remove the public IP addresses, you are told those VMs need to communicate to external sites as part of the customer's typical operations.

What should you recommend to reduce the need for public IP addresses in your customer's VMs?

  1. Google Cloud Armor
  2. Cloud NAT
  3. Cloud Router
  4. Cloud VPN
Show answer and explanation

Correct answer: B. Cloud NAT

Cloud NAT allows VMs without public IP addresses to initiate outbound connections to external sites by translating their private IPs to the NAT gateway's IP address. This eliminates the need for public IPs on individual VMs while maintaining their ability to communicate with external resources. Cloud Armor provides DDoS protection but doesn't address IP address management. Cloud Router manages BGP routing. Cloud VPN creates encrypted tunnels but doesn't reduce the need for public IPs for external communication.

Why the other options are wrong

  • A. Cloud Armor is a DDoS and application protection service, not a solution for managing outbound traffic without public IPs.
  • C. Cloud Router manages dynamic routing via BGP and does not provide the ability to hide internal IPs from external services.
  • D. Cloud VPN creates encrypted tunnels between networks but does not eliminate the need for public IPs for general external communication.

Question 61

Your company's new CEO recently sold two of the company's divisions. Your Director asks you to help migrate the Google Cloud projects associated with those divisions to a new organization node.

Which preparation steps are necessary before this migration occurs? (Choose two.)

  1. Remove all project-level custom Identity and Access Management (IAM) roles.
  2. Disallow inheritance of organization policies.
  3. Identify inherited Identity and Access Management (IAM) roles on projects to be migrated.
  4. Create a new folder for all projects to be migrated.
  5. Remove the specific migration projects from any VPC Service Controls perimeters and bridges.
Show answer and explanation

Correct answer: C, E. Identify inherited Identity and Access Management (IAM) roles on projects to be migrated; Remove the specific migration projects from any VPC Service Controls perimeters and bridges.

Before migrating projects to a new organization, inherited IAM roles must be identified to understand what permissions will be lost or changed in the new organization, as inheritance rules differ between organizations. Additionally, projects must be removed from any VPC Service Controls perimeters and bridges before migration, as these security perimeters cannot span multiple organizations and would block access post-migration.

Why the other options are wrong

  • A. Custom IAM roles at the project level do not prevent migration and can be recreated in the new organization if needed.
  • B. Organization policies are automatically re-evaluated in the new organization; disallowing inheritance beforehand creates unnecessary complexity.
  • D. Creating a folder for projects to be migrated is not a necessary preparation step; projects can be organized in the new organization after migration.

Question 62

A service account key has been publicly exposed on multiple public code repositories. After reviewing the logs, you notice that the keys were used to generate short-lived credentials. You need to immediately remove access with the service account.

What should you do?

  1. Delete the compromised service account.
  2. Disable the compromised service account key.
  3. Wait until the service account credentials expire automatically.
  4. Rotate the compromised service account key.
Show answer and explanation

Correct answer: B. Disable the compromised service account key.

When a service account key is publicly exposed and used maliciously, the immediate action is to disable that specific key. Disabling the key revokes access immediately without deleting the entire service account, which may still be needed for legitimate operations. Short-lived credentials generated from the compromised key will also become invalid once the source key is disabled. This is faster and more targeted than waiting for credential expiration.

Why the other options are wrong

  • A. Deleting the entire service account is unnecessarily destructive and may break legitimate services still using that account.
  • C. Waiting for automatic expiration is too slow when a key has been publicly exposed and is actively being misused.
  • D. Rotating the key does not immediately revoke the compromised key's ability to generate new credentials.

Question 63

You want to prevent users from accidentally deleting a Shared VPC host project.

Which organization-level policy constraint should you enable?

  1. compute.restrictSharedVpcHostProjects
  2. compute.restrictXpnProjectLienRemoval
  3. compute.restrictSharedVpcSubnetworks
  4. compute.sharedReservationsOwnerProjects
Show answer and explanation

Correct answer: B. compute.restrictXpnProjectLienRemoval

The compute.restrictXpnProjectLienRemoval constraint prevents removal of the Shared VPC host project lien, which protects against accidental deletion. A lien is an organizational policy enforcement mechanism that blocks project deletion. This is the specific constraint designed to prevent users from accidentally deleting a Shared VPC host project at the organization level.

Why the other options are wrong

  • A. compute.restrictSharedVpcHostProjects restricts which projects can be designated as Shared VPC hosts, not preventing deletion of existing hosts.
  • C. compute.restrictSharedVpcSubnetworks restricts which subnetworks can be attached to Shared VPC, not preventing project deletion.
  • D. compute.sharedReservationsOwnerProjects relates to shared commitment reservations, not Shared VPC host project protection.

Question 64

You are creating a new infrastructure CI/CD pipeline to deploy hundreds of ephemeral projects in your Google Cloud organization to enable your users to interact with Google Cloud. You want to restrict the use of the default networks in your organization while following Google-recommended best practices.

What should you do?

  1. Enable the constraints/compute.skipDefaultNetworkCreation organization policy constraint at the organization level.
  2. Create a cron job to trigger a daily Cloud Function to automatically delete all default networks for each project.
  3. Grant your users the IAM Owner role at the organization level. Create a VPC Service Controls perimeter around the project that restricts the compute.googleapis.com API.
  4. Only allow your users to use your CI/CD pipeline with a predefined set of infrastructure templates they can deploy to skip the creation of the default networks.
Show answer and explanation

Correct answer: A. Enable the constraints/compute.skipDefaultNetworkCreation organization policy constraint at the organization level.

The constraints/compute.skipDefaultNetworkCreation organization policy constraint is the Google-recommended best practice for preventing default networks from being created in new projects. This constraint should be enabled at the organization level to automatically prevent default network creation across all projects, ensuring compliance with the requirement to restrict default network usage. A cron job approach is not scalable, granting Owner role violates least privilege principles, and relying solely on CI/CD templates does not prevent direct project creation outside the pipeline.

Why the other options are wrong

  • B. A cron job approach is inefficient, reactive, and does not scale well for hundreds of ephemeral projects compared to proactive organization policy enforcement.
  • C. Granting IAM Owner role violates least privilege principles and does not prevent default network creation as effectively as an organization policy constraint.
  • D. Relying only on CI/CD templates does not prevent users from creating projects outside the pipeline or manually creating default networks within projects.

Question 65

In a shared security responsibility model for IaaS, which two layers of the stack does the customer share responsibility for? (Choose two.)

  1. Hardware
  2. Network Security
  3. Storage Encryption
  4. Access Policies
  5. Boot
Show answer and explanation

Correct answer: B, D. Network Security; Access Policies

In an Infrastructure-as-a-Service (IaaS) shared security responsibility model, the customer shares responsibility for Network Security (configuring firewalls, security groups, network policies) and Access Policies (controlling who can access resources, IAM configurations). The cloud provider manages hardware, boot, and the hypervisor, while customers control the network and access layers above the infrastructure.

Why the other options are wrong

  • A. Hardware is the responsibility of the cloud provider in IaaS; customers do not manage physical infrastructure.
  • C. Storage Encryption at the infrastructure level is primarily the cloud provider's responsibility, though customers may add application-level encryption.
  • E. Boot is part of the underlying infrastructure managed by the cloud provider, not shared customer responsibility.

Question 66

As adoption of the Cloud Data Loss Prevention (Cloud DLP) API grows within your company, you need to optimize usage to reduce cost. Cloud DLP target data is stored in Cloud Storage and BigQuery. The location and region are identified as a suffix in the resource name.

Which cost reduction options should you recommend?

  1. Set appropriate rowsLimit value on BigQuery data hosted outside the US and set appropriate bytesLimitPerFile value on multiregional Cloud Storage buckets.
  2. Set appropriate rowsLimit value on BigQuery data hosted outside the US, and minimize transformation units on multiregional Cloud Storage buckets.
  3. Use rowsLimit and bytesLimitPerFile to sample data and use CloudStorageRegexFileSet to limit scans.
  4. Use FindingLimits and TimespanContfig to sample data and minimize transformation units.
Show answer and explanation

Correct answer: C. Use rowsLimit and bytesLimitPerFile to sample data and use CloudStorageRegexFileSet to limit scans.

Cost reduction is achieved by using rowsLimit and bytesLimitPerFile to sample data rather than scanning entire datasets, and using CloudStorageRegexFileSet to limit the scope of file scans to specific patterns or locations. This combination minimizes data processed and therefore reduces transformation units and API costs.

Why the other options are wrong

  • A. While rowsLimit and bytesLimitPerFile are useful, this option does not mention CloudStorageRegexFileSet which is essential for limiting scan scope, and it focuses only on specific data locations rather than general optimization.
  • B. This option mentions rowsLimit but transformation units cannot be directly minimized independently; they depend on the volume of data scanned.
  • D. FindingLimits and TimespanConfig are not primary cost optimization mechanisms; sampling through rowsLimit and bytesLimitPerFile with targeted scans are more effective.

Question 67

You are a security engineer at a finance company. Your organization plans to store data on Google Cloud, but your leadership team is worried about the security of their highly sensitive data. Specifically, your company is concerned about internal Google employees' ability to access your company's data on Google Cloud.

What solution should you propose?

  1. Use customer-managed encryption keys.
  2. Use Google's Identity and Access Management (IAM) service to manage access controls on Google Cloud.
  3. Enable Admin activity logs to monitor access to resources.
  4. Enable Access Transparency logs with Access Approval requests for Google employees.
Show answer and explanation

Correct answer: D. Enable Access Transparency logs with Access Approval requests for Google employees.

Access Transparency logs with Access Approval requests directly addresses the concern about internal Google employee access. Access Approval requires Google employees to request and justify access to customer data, with customers receiving notifications and the ability to approve or deny each request. This gives the finance company explicit visibility and control over Google employee access to their sensitive data, which is the specific concern raised by leadership.

Why the other options are wrong

  • A. Customer-managed encryption keys prevent Google from reading data but do not address or prevent Google employees from requesting access.
  • B. IAM access controls manage which Google Cloud users can access resources but do not prevent or monitor Google employees' access.
  • C. Admin activity logs show what actions occurred but do not prevent or require approval for Google employee access to customer data.

Question 68

You want to use the gcloud command-line tool to authenticate using a third-party single sign-on (SSO) SAML identity provider.

Which options are necessary to ensure that authentication is supported by the third-party identity provider (IdP)? (Choose two.)

  1. SSO SAML as a third-party IdP
  2. Identity Platform
  3. OpenID Connect
  4. Identity-Aware Proxy
  5. Cloud Identity
Show answer and explanation

Correct answer: A, E. SSO SAML as a third-party IdP; Cloud Identity

To authenticate using a third-party SAML identity provider with gcloud, you need the IdP to support SAML as a third-party identity provider (option A), and you need Google Cloud Identity to manage and configure the SAML integration (option E). Cloud Identity is the Google Cloud service that handles identity federation and SAML configuration. The IdP must explicitly support SAML federation with Google Cloud for this authentication flow to work.

Why the other options are wrong

  • B. Identity Platform is used for application-level authentication but is not the required service for gcloud CLI SAML authentication.
  • C. OpenID Connect is an alternative authentication protocol but is not necessary for SAML-based authentication.
  • D. Identity-Aware Proxy protects access to applications but is not required for SAML identity provider authentication.

Question 69

Your team wants to make sure Compute Engine instances running in your production project do not have public IP addresses. The frontend application Compute Engine instances will require public IPs. The product engineers have the Editor role to modify resources. Your team wants to enforce this requirement.

How should your team meet these requirements?

  1. Enable Private Access on the VPC network in the production project.
  2. Remove the Editor role and grant the Compute Admin IAM role to the engineers.
  3. Set up an organization policy to only permit public IPs for the front-end Compute Engine instances.
  4. Set up a VPC network with two subnets: one with public IPs and one without public IPs.
Show answer and explanation

Correct answer: C. Set up an organization policy to only permit public IPs for the front-end Compute Engine instances.

Organization policies are the correct mechanism to enforce constraints on resource creation across a project. By setting up an organization policy that restricts public IP assignment while creating exceptions for specific instances (the front-end ones), the team can enforce the requirement that most instances lack public IPs while allowing frontend instances to have them. This prevents engineers from circumventing the policy even with Editor role permissions.

Why the other options are wrong

  • A. Private Access enables private connectivity to Google services but does not prevent public IP assignment to instances.
  • B. Removing Editor role and granting Compute Admin would still allow those engineers to create public IPs; IAM roles do not enforce this type of resource constraint.
  • D. VPC subnet configuration alone cannot enforce a policy, subnets can be configured for public or private IPs, but instances can still be assigned public IPs in either case through explicit configuration.

Question 70

Your team needs to make sure that their backend database can only be accessed by the frontend application and no other instances on the network.

How should your team design this network?

  1. Create an ingress firewall rule to allow access only from the application to the database using firewall tags.
  2. Create a different subnet for the frontend application and database to ensure network isolation.
  3. Create two VPC networks, and connect the two networks using Cloud VPN gateways to ensure network isolation.
  4. Create two VPC networks, and connect the two networks using VPC peering to ensure network isolation.
Show answer and explanation

Correct answer: A. Create an ingress firewall rule to allow access only from the application to the database using firewall tags.

Creating an ingress firewall rule using firewall tags to allow access only from the frontend application to the database is the correct approach. Firewall tags enable granular, identity-based network access control within a single VPC. This method isolates database traffic to only authorized application instances without requiring multiple subnets or networks, and is more efficient and manageable than network-level segmentation.

Why the other options are wrong

  • B. Subnets provide network routing boundaries but not access control; instances in different subnets can still communicate without firewall rules to restrict it.
  • C. Using separate VPC networks connected by Cloud VPN adds unnecessary complexity and latency for what can be accomplished with firewall rules.
  • D. VPC peering is used for connecting entire networks, not for restricting access between specific application tiers within a single organization.

Question 71

A customer wants to deploy a large number of 3-tier web applications on Compute Engine.

How should the customer ensure authenticated network separation between the different tiers of the application?

  1. Run each tier in its own Project, and segregate using Project labels.
  2. Run each tier with a different Service Account (SA), and use SA-based firewall rules.
  3. Run each tier in its own subnet, and use subnet-based firewall rules.
  4. Run each tier with its own VM tags, and use tag-based firewall rules.
Show answer and explanation

Correct answer: B. Run each tier with a different Service Account (SA), and use SA-based firewall rules.

Using different Service Accounts for each tier and creating firewall rules based on Service Account identity provides authenticated network separation. Service Account-based firewall rules enforce identity-aware access control, ensuring only the appropriate service accounts can communicate between tiers. This is more granular and maintainable than tag-based rules and is the recommended practice for tier-to-tier segmentation.

Why the other options are wrong

  • A. Separating tiers into different projects is unnecessarily complex and doesn't provide network-level separation; project labels are metadata only.
  • C. Subnet-based firewall rules can work but are less flexible than service account-based rules and don't provide true authentication; service accounts are the recommended approach.
  • D. While tag-based firewall rules can provide separation, they lack the authenticated identity verification that service accounts provide, making them less secure for compliance requirements.

Question 72

In an effort for your company messaging app to comply with FIPS 140-2, a decision was made to use GCP compute and network services. The messaging app architecture includes a Managed Instance Group (MIG) that controls a cluster of Compute Engine instances. The instances use Local SSDs for data caching and UDP for instance-to-instance communications. The app development team is willing to make any changes necessary to comply with the standard

Which options should you recommend to meet the requirements?

  1. Encrypt all cache storage and VM-to-VM communication using the BoringCrypto module.
  2. Set Disk Encryption on the Instance Template used by the MIG to customer-managed key and use BoringSSL for all data transit between instances.
  3. Change the app instance-to-instance communications from UDP to TCP and enable BoringSSL on clients' TLS connections.
  4. Set Disk Encryption on the Instance Template used by the MIG to Google-managed Key and use BoringSSL library on all instance-to-instance communications.
Show answer and explanation

Correct answer: D. Set Disk Encryption on the Instance Template used by the MIG to Google-managed Key and use BoringSSL library on all instance-to-instance communications.

FIPS 140-2 compliance requires encryption at rest and in transit using FIPS-validated cryptographic modules. Google-managed keys with Cloud Encryption provide FIPS compliance for disk encryption on Compute Engine instances. BoringSSL (the FIPS-capable variant used by Google) must be used for VM-to-VM communication encryption, making it compliant regardless of whether UDP or TCP is used. Google-managed encryption is FIPS-validated and simpler than customer-managed key rotation.

Why the other options are wrong

  • A. BoringCrypto module exists but is not the complete solution; it does not address the specific requirement that encryption keys must be FIPS-validated, and Google-managed keys are preferred for compliance.
  • B. Customer-managed keys add complexity and responsibility for key management; Google-managed keys are already FIPS-validated, and BoringSSL alone does not guarantee FIPS compliance without proper key management.
  • C. Changing UDP to TCP is unnecessary for FIPS compliance; the protocol layer does not determine FIPS status. BoringSSL alone without proper key management does not ensure FIPS compliance.

Question 73

Which Identity-Aware Proxy role should you grant to an Identity and Access Management (IAM) user to access HTTPS resources?

  1. Security Reviewer
  2. IAP-Secured Tunnel User
  3. IAP-Secured Web App User
  4. Service Broker Operator
Show answer and explanation

Correct answer: C. IAP-Secured Web App User

The IAP-Secured Web App User role is the correct role to grant to IAM users who need to access HTTPS resources protected by Identity-Aware Proxy. This role allows users to access web applications secured by IAP without needing direct network access. Security Reviewer is for auditing purposes, IAP-Secured Tunnel User is for accessing resources via SSH/RDP tunnels, and Service Broker Operator is unrelated to IAP resource access.

Why the other options are wrong

  • A. Security Reviewer role is designed for auditing and monitoring IAP access, not for users who need to access HTTPS resources.
  • B. IAP-Secured Tunnel User role is for accessing non-HTTPS resources through SSH/RDP tunnels, not for HTTPS web applications.
  • D. Service Broker Operator role is unrelated to Identity-Aware Proxy and does not grant access to IAP-secured resources.

Question 74

You are asked to recommend a solution to store and retrieve sensitive configuration data from an application that runs on Compute Engine.

Which option should you recommend?

  1. Cloud Key Management Service
  2. Compute Engine guest attributes
  3. Compute Engine custom metadata
  4. Secret Manager
Show answer and explanation

Correct answer: D. Secret Manager

Secret Manager is the purpose-built Google Cloud service for storing, managing, and retrieving sensitive configuration data like API keys, passwords, and credentials. It provides encryption at rest, automatic rotation, audit logging, and fine-grained IAM access control. Cloud KMS manages encryption keys themselves, guest attributes and custom metadata are not designed for sensitive data security.

Why the other options are wrong

  • A. Cloud KMS is for managing encryption keys, not for storing application secrets and configuration data.
  • B. Guest attributes are intended for VM configuration and metadata sharing, not sensitive secret storage.
  • C. Custom metadata is not encrypted by default and lacks the security controls needed for sensitive data.

Question 75

You have an application where the frontend is deployed on a managed instance group in subnet A and the data layer is stored on a mysql Compute Engine virtual machine (VM) in subnet B on the same VPC. Subnet A and Subnet B hold several other Compute Engine VMs. You only want to allow the application frontend to access the data in the application's mysql instance on port 3306.

What should you do?

  1. Configure an ingress firewall rule that allows communication from the src IP range of subnet A to the tag "data-tag" that is applied to the mysql Compute Engine VM on port 3306.
  2. Configure an ingress firewall rule that allows communication from the frontend's unique service account to the unique service account of the mysql Compute Engine VM on port 3306.
  3. Configure a network tag "fe-tag" to be applied to all instances in subnet A and a network tag "data-tag" to be applied to all instances in subnet B. Then configure an egress firewall rule that allows communication from Compute Engine VMs tagged with data-tag to destination Compute Engine VMs tagged fe-tag.
  4. Configure a network tag "fe-tag" to be applied to all instances in subnet A and a network tag "data-tag" to be applied to all instances in subnet B. Then configure an ingress firewall rule that allows communication from Compute Engine VMs tagged with fe-tag to destination Compute Engine VMs tagged with data-tag.
Show answer and explanation

Correct answer: B. Configure an ingress firewall rule that allows communication from the frontend's unique service account to the unique service account of the mysql Compute Engine VM on port 3306.

Configuring an ingress firewall rule that identifies sources and destinations by service account is the most precise and scalable approach. This allows the frontend application's unique service account to access the mysql VM's service account on port 3306, maintaining security through identity-based access control rather than IP-based rules.

Why the other options are wrong

  • A. Using subnet IP ranges is overly broad as it would allow any VM in subnet A to access the mysql instance, not just the application frontend.
  • C. This creates an egress rule from data-tag to fe-tag, which is backwards; it should allow ingress to the mysql instance, not egress from it.
  • D. While this uses the correct tags and direction, applying tags to all instances in a subnet is less granular than using service accounts; it would allow unintended VMs in subnet A to access the mysql instance.

75 questions in. Want the rest? The first half of the pack, free by email · The full pack

Question 76

An organization is migrating from their current on-premises productivity software systems to G Suite. Some network security controls were in place that were mandated by a regulatory body in their region for their previous on-premises system. The organization's risk team wants to ensure that network security controls are maintained and effective in G Suite. A security architect supporting this migration has been asked to ensure that network security controls are in place as part of the new shared responsibility model between the organization and Google Cloud.

What solution would help meet the requirements?

  1. Ensure that firewall rules are in place to meet the required controls.
  2. Set up Cloud Armor to ensure that network security controls can be managed for G Suite.
  3. Network security is a built-in solution and Google's Cloud responsibility for SaaS products like G Suite.
  4. Set up an array of Virtual Private Cloud (VPC) networks to control network security as mandated by the relevant regulation.
Show answer and explanation

Correct answer: C. Network security is a built-in solution and Google's Cloud responsibility for SaaS products like G Suite.

For SaaS products like G Suite, network security is Google's responsibility, not the customer's. The shared responsibility model for SaaS means Google handles infrastructure security including network controls. The organization cannot implement network-level firewall rules or VPC controls for G Suite because it operates outside their infrastructure. Instead, they should rely on G Suite's built-in security controls and compliance certifications to meet regulatory requirements.

Why the other options are wrong

  • A. Firewall rules apply to GCP infrastructure, not to G Suite SaaS services accessed over the internet.
  • B. Cloud Armor protects GCP resources from DDoS and application-layer attacks; it cannot control G Suite network security.
  • D. VPC networks control traffic within GCP infrastructure and cannot be applied to manage G Suite security, which is a Google-hosted service.

Question 77

You have been tasked with implementing external web application protection against common web application attacks for a public application on Google Cloud.

You want to validate these policy changes before they are enforced.

What service should you use?

  1. Google Cloud Armor's preconfigured rules in preview mode
  2. Prepopulated VPC firewall rules in monitor mode
  3. The inherent protections of Google Front End (GFE)
  4. Cloud Load Balancing firewall rules
  5. VPC Service Controls in dry run mode
Show answer and explanation

Correct answer: A. Google Cloud Armor's preconfigured rules in preview mode

Google Cloud Armor's preconfigured rules in preview mode allows you to validate policy changes before enforcing them. Preview mode lets you test rules against real traffic to see their impact without blocking requests, making it ideal for validating changes before enforcement. This is the only option that provides both protection against common web attacks and a validation/testing mechanism.

Why the other options are wrong

  • B. VPC firewall rules in monitor mode protect at the network layer, not the application layer where web attacks occur.
  • C. GFE provides inherent protections but offers no mechanism to validate custom policy changes before enforcement.
  • D. Cloud Load Balancing firewall rules do not provide the same validation testing capability as Cloud Armor preview mode.
  • E. VPC Service Controls in dry run mode protects data access, not web applications against web-layer attacks.

Question 78

You're developing the incident response plan for your company. You need to define the access strategy that your DevOps team will use when reviewing and investigating a deployment issue in your Google Cloud environment. There are two main requirements:

• Least-privilege access must be enforced at all times.

• The DevOps team must be able to access the required resources only during the deployment issue.

How should you grant access while following Google-recommended best practices?

  1. Assign the Project Viewer Identity and Access Management (IAM) role to the DevOps team.
  2. Create a custom IAM role with limited list/view permissions, and assign it to the DevOps team.
  3. Create a service account, and grant it the Project Owner IAM role. Give the Service Account User Role on this service account to the DevOps team.
  4. Create a service account, and grant it limited list/view permissions. Give the Service Account User Role on this service account to the DevOps team.
Show answer and explanation

Correct answer: D. Create a service account, and grant it limited list/view permissions. Give the Service Account User Role on this service account to the DevOps team.

The recommended approach combines a service account with limited permissions (least-privilege principle) and the Service Account User role granted to the DevOps team. This allows temporary elevated access only when needed for investigation while maintaining strict privilege separation. The service account should have only the specific list/view permissions required for investigating deployment issues, not broad roles like Project Owner or Project Viewer.

Why the other options are wrong

  • A. Project Viewer grants broad read access across all resources, violating least-privilege principles and providing more access than necessary for investigating a specific deployment issue.
  • B. Custom IAM roles with view-only permissions do not provide the temporary access control and audit trail that service accounts enable, and still represent standing access rather than time-limited incident response access.
  • C. Granting Project Owner role to a service account violates least-privilege principles and provides far more permissions than needed to investigate a deployment issue.

Question 79

A large financial institution is moving its Big Data analytics to Google Cloud Platform. They want to have maximum control over the encryption process of data stored at rest in BigQuery.

What technique should the institution use?

  1. Use Cloud Storage as a federated Data Source.
  2. Use a Cloud Hardware Security Module (Cloud HSM).
  3. Customer-managed encryption keys (CMEK).
  4. Customer-supplied encryption keys (CSEK).
Show answer and explanation

Correct answer: C. Customer-managed encryption keys (CMEK).

BigQuery supports customer-managed encryption keys held in Cloud KMS, which lets the customer control key creation, rotation, disabling and destruction for data at rest. Customer-supplied encryption keys are not supported by BigQuery, so CMEK is the maximum level of control available.

Why the other options are wrong

  • A. Using Cloud Storage as a federated source does not give control over BigQuery's own encryption at rest.
  • B. Cloud HSM is a protection level for Cloud KMS keys, and it is used through CMEK rather than being a BigQuery encryption technique on its own.
  • D. BigQuery does not support customer-supplied encryption keys.

Question 80

You want to make sure that your organization's Cloud Storage buckets cannot have data publicly available to the internet. You want to enforce this across all Cloud Storage buckets.

What should you do?

  1. Remove Owner roles from end users, and configure Cloud Data Loss Prevention.
  2. Remove Owner roles from end users, and enforce domain restricted sharing in an organization policy.
  3. Configure uniform bucket-level access, and enforce domain restricted sharing in an organization policy.
  4. Remove *.setIamPolicy permissions from all roles, and enforce domain restricted sharing in an organization policy.
Show answer and explanation

Correct answer: C. Configure uniform bucket-level access, and enforce domain restricted sharing in an organization policy.

To prevent public data access across all Cloud Storage buckets, you should configure uniform bucket-level access to standardize access control at the bucket level, and enforce domain restricted sharing through an organization policy to prevent sharing with public identities. This combination prevents both ACL-based public access and IAM-based public sharing across the organization.

Why the other options are wrong

  • A. Removing Owner roles and Cloud Data Loss Prevention does not directly prevent public bucket access; DLP is for data content scanning, not access control.
  • B. Removing Owner roles is overly restrictive and does not address the core issue; domain restricted sharing alone without uniform bucket-level access is insufficient.
  • D. Removing setIamPolicy permissions would break normal access management; domain restricted sharing is the proper organizational policy control.

Question 81

You are troubleshooting access denied errors between Compute Engine instances connected to a Shared VPC and BigQuery datasets. The datasets reside in a project protected by a VPC Service Controls perimeter.

What should you do?

  1. Add the host project containing the Shared VPC to the service perimeter.
  2. Add the service project where the Compute Engine instances reside to the service perimeter.
  3. Create a service perimeter between the service project where the Compute Engine instances reside and the host project that contains the Shared VPC.
  4. Create a perimeter bridge between the service project where the Compute Engine instances reside and the perimeter that contains the protected BigQuery datasets.
Show answer and explanation

Correct answer: A. Add the host project containing the Shared VPC to the service perimeter.

When a VM uses a Shared VPC network, VPC Service Controls treats the request as coming from the host project that owns the network. The host project must therefore be added to the service perimeter protecting the BigQuery datasets for the requests to be allowed.

Why the other options are wrong

  • B. Adding only the service project is not enough, because the network traffic is attributed to the Shared VPC host project.
  • C. A perimeter around the service and host projects alone does not include the BigQuery project, so access is still blocked.
  • D. A perimeter bridge with the service project does not help because the requests are evaluated against the host project's network.

Question 82

A company's application is deployed with a user-managed Service Account key. You want to use Google-recommended practices to rotate the key.

What should you do?

  1. Open Cloud Shell and run gcloud iam service-accounts enable-auto-rotate --iam-account=IAM_ACCOUNT.
  2. Open Cloud Shell and run gcloud iam service-accounts keys rotate --iam-account=IAM_ACCOUNT --key=NEW_KEY.
  3. Create a new key, and use the new key in the application. Delete the old key from the Service Account.
  4. Create a new key, and use the new key in the application. Store the old key on the system as a backup key.
Show answer and explanation

Correct answer: C. Create a new key, and use the new key in the application. Delete the old key from the Service Account.

Google-recommended practices for rotating user-managed Service Account keys require creating a new key and updating the application to use it before deleting the old key. This ensures zero-downtime rotation and allows time to verify the new key works correctly. The old key should be deleted only after confirming the new key is functioning properly in the application, not retained as a backup.

Why the other options are wrong

  • A. The gcloud command shown is not a valid Cloud IAM command for key rotation.
  • B. The gcloud command shown does not exist and is not the correct way to rotate Service Account keys.
  • D. Storing the old key as a backup increases the attack surface and violates the principle of minimizing active credentials in production systems.

Question 83

Your company operates an application instance group that is currently deployed behind a Google Cloud load balancer in us-central-1 and is configured to use the Standard Tier network. The infrastructure team wants to expand to a second Google Cloud region, us-east-2. You need to set up a single external IP address to distribute new requests to the instance groups in both regions.

What should you do?

  1. Change the load balancer backend configuration to use network endpoint groups instead of instance groups.
  2. Change the load balancer frontend configuration to use the Premium Tier network, and add the new instance group.
  3. Create a new load balancer in us-east-2 using the Standard Tier network, and assign a static external IP address.
  4. Create a Cloud VPN connection between the two regions, and enable Google Private Access.
Show answer and explanation

Correct answer: B. Change the load balancer frontend configuration to use the Premium Tier network, and add the new instance group.

To distribute requests to instance groups across multiple regions with a single external IP address, the load balancer must use the Premium Tier network service level. Standard Tier only supports single-region backends, while Premium Tier enables cross-region load balancing with a single frontend configuration and IP address spanning us-central-1 and us-east-2.

Why the other options are wrong

  • A. Network endpoint groups are a backend configuration option but do not enable multi-region load balancing with a single IP address if the tier remains Standard.
  • C. Creating a separate load balancer in the second region would result in multiple external IP addresses, not the single IP address requirement.
  • D. Cloud VPN and Private Google Access address network connectivity and private API access, not cross-region load balancing with a single frontend IP.

Question 84

You work for an organization in a regulated industry that has strict data protection requirements. The organization backs up their data in the cloud. To comply with data privacy regulations, this data can only be stored for a specific length of time and must be deleted after this specific period.

You want to automate the compliance with this regulation while minimizing storage costs.

What should you do?

  1. Store the data in a persistent disk, and delete the disk at expiration time.
  2. Store the data in a Cloud Bigtable table, and set an expiration time on the column families.
  3. Store the data in a BigQuery table, and set the table's expiration time.
  4. Store the data in a Cloud Storage bucket, and configure the bucket's Object Lifecycle Management feature.
Show answer and explanation

Correct answer: D. Store the data in a Cloud Storage bucket, and configure the bucket's Object Lifecycle Management feature.

Cloud Storage with Object Lifecycle Management is the most cost-effective solution for automatically deleting backup data after a specific period. Lifecycle policies can be configured to delete objects based on age, automating compliance with data retention regulations while minimizing storage costs by removing expired data without manual intervention.

Why the other options are wrong

  • A. Persistent disks require manual deletion and are designed for active compute instances, not long-term backup storage, making them inefficient and expensive.
  • B. While Cloud Bigtable column families support TTL expiration, Bigtable is optimized for high-throughput analytics and time-series data, not cost-effective backup storage.
  • C. BigQuery table expiration can delete tables automatically, but BigQuery is optimized for analytics queries and is more expensive than Cloud Storage for simple backup retention.

Question 85

You plan to deploy your cloud infrastructure using a CI/CD cluster hosted on Compute Engine. You want to minimize the risk of its credentials being stolen by a third party.

What should you do?

  1. Create a dedicated Cloud Identity user account for the cluster. Use a strong self-hosted vault solution to store the user's temporary credentials.
  2. Create a dedicated Cloud Identity user account for the cluster. Enable the constraints/iam.disableServiceAccountCreation organization policy at the project level.
  3. Create a custom service account for the cluster. Enable the constraints/iam.disableServiceAccountKeyCreation organization policy at the project level
  4. Create a custom service account for the cluster. Enable the constraints/iam.allowServiceAccountCredentialLifetimeExtension organization policy at the project level.
Show answer and explanation

Correct answer: C. Create a custom service account for the cluster. Enable the constraints/iam.disableServiceAccountKeyCreation organization policy at the project level

For a CI/CD cluster on Compute Engine, a custom service account should be created to follow the principle of least privilege. The constraints/iam.disableServiceAccountKeyCreation organization policy should be enabled at the project level to prevent the creation of service account keys, eliminating the risk of key theft. This forces the use of identity federation and temporary credentials instead of long-lived keys.

Why the other options are wrong

  • A. Cloud Identity user accounts are not appropriate for service-to-service authentication; service accounts should be used instead.
  • B. Disabling service account creation would prevent creating the dedicated service account needed for the cluster.
  • D. The allowServiceAccountCredentialLifetimeExtension policy extends credential lifetime, increasing security risk rather than minimizing it.

Question 86

You need to create a VPC that enables your security team to control network resources such as firewall rules.

How should you configure the network to allow for separation of duties for network resources?

  1. Set up multiple VPC networks, and set up multi-NIC virtual appliances to connect the networks.
  2. Set up VPC Network Peering, and allow developers to peer their network with a Shared VPC.
  3. Set up a VPC in a project. Assign the Compute Network Admin role to the security team, and assign the Compute Admin role to the developers.
  4. Set up a Shared VPC where the security team manages the firewall rules, and share the network with developers via service projects.
Show answer and explanation

Correct answer: D. Set up a Shared VPC where the security team manages the firewall rules, and share the network with developers via service projects.

Shared VPC enables separation of duties by allowing the security team to manage network resources (firewall rules) in a host project while developers deploy resources in service projects. This architecture provides centralized network control without requiring developers to have network administration privileges. Multiple VPCs with appliances creates operational complexity, VPC Peering does not provide centralized control, and assigning roles in a single VPC does not achieve the separation of duties needed for multi-team environments.

Why the other options are wrong

  • A. Multiple VPCs with multi-NIC appliances creates operational complexity and is not the recommended approach for achieving separation of duties.
  • B. VPC Peering allows connection between networks but does not provide centralized firewall rule management or separation of duties between teams.
  • C. Assigning different roles in a single VPC does not achieve separation of duties as effectively as Shared VPC, which isolates network administration from resource deployment.

Question 87

You are a security administrator at your company. Per Google-recommended best practices, you implemented the domain restricted sharing organization policy to allow only required domains to access your projects. An engineering team is now reporting that users at an external partner outside your organization domain cannot be granted access to the resources in a project.

How should you make an exception for your partner's domain while following the stated best practices?

  1. Turn off the domain restriction sharing organization policy. Set the policy value to "Allow All."
  2. Turn off the domain restricted sharing organization policy. Provide the external partners with the required permissions using Google's Identity and Access Management (IAM) service.
  3. Turn off the domain restricted sharing organization policy. Add each partner's Google Workspace customer ID to a Google group, add the Google group as an exception under the organization policy, and then turn the policy back on.
  4. Turn off the domain restricted sharing organization policy. Set the policy value to "Custom." Add each external partner's Cloud Identity or Google Workspace customer ID as an exception under the organization policy, and then turn the policy back on.
Show answer and explanation

Correct answer: D. Turn off the domain restricted sharing organization policy. Set the policy value to "Custom." Add each external partner's Cloud Identity or Google Workspace customer ID as an exception under the organization policy, and then turn the policy back on.

When using domain restricted sharing organization policies, you can configure exceptions by setting the policy value to "Custom" and adding specific external partner Cloud Identity or Google Workspace customer IDs as exceptions. This maintains the security posture of the organization policy while allowing controlled access to specific trusted external domains. The policy must be temporarily disabled to make these configuration changes, then re-enabled with the custom exceptions in place.

Why the other options are wrong

  • A. Turning off the policy entirely and allowing all domains defeats the purpose of implementing domain restriction for security.
  • B. Simply disabling the policy and using IAM alone removes the organizational control enforced by the policy and doesn't follow the stated best practice of maintaining domain restrictions.
  • C. While Google groups can be used for other purposes, the correct approach for organization policies is to add customer IDs directly as exceptions, not through group membership.

Question 88

An employer wants to track how bonus compensations have changed over time to identify employee outliers and correct earning disparities. This task must be performed without exposing the sensitive compensation data for any individual and must be reversible to identify the outlier.

Which Cloud Data Loss Prevention API technique should you use to accomplish this?

  1. Generalization
  2. Redaction
  3. CryptoHashConfig
  4. CryptoReplaceFfxFpeConfig
Show answer and explanation

Correct answer: D. CryptoReplaceFfxFpeConfig

The data must be de-identified yet reversible so an outlier can be re-identified later. CryptoReplaceFfxFpeConfig applies format-preserving encryption with a key, so values can be decrypted to recover the original record when needed. Hashing, generalization and redaction cannot be reversed.

Why the other options are wrong

  • A. Generalization replaces values with ranges or broader categories and cannot be reversed to the original value.
  • B. Redaction removes the data entirely, so it cannot be recovered.
  • C. CryptoHashConfig produces a one-way hash, so the original value cannot be restored.

Question 89

For compliance reasons, an organization needs to ensure that in-scope PCI Kubernetes Pods reside on `in-scope` Nodes only. These Nodes can only contain the `in-scope` Pods.

How should the organization achieve this objective?

  1. Add a nodeSelector field to the pod configuration to only use the Nodes labeled inscope: true.
  2. Create a node pool with the label inscope: true and a Pod Security Policy that only allows the Pods to run on Nodes with that label.
  3. Place a taint on the Nodes with the label inscope: true and effect NoSchedule and a toleration to match in the Pod configuration.
  4. Run all in-scope Pods in the namespace "in-scope-pci".
Show answer and explanation

Correct answer: C. Place a taint on the Nodes with the label inscope: true and effect NoSchedule and a toleration to match in the Pod configuration.

Taints and tolerations provide the enforcement mechanism needed for PCI compliance. By placing a taint with NoSchedule effect on in-scope nodes, only pods with matching tolerations can be scheduled there. This ensures strict isolation: in-scope pods run only on in-scope nodes, and in-scope nodes run only in-scope pods. This is the proper Kubernetes pattern for such compliance requirements.

Why the other options are wrong

  • A. nodeSelector only controls where pods can run but does not prevent non-in-scope pods from also using those nodes.
  • B. Pod Security Policies control pod-level security settings but do not enforce node affinity; they cannot ensure that only in-scope pods run on in-scope nodes.
  • D. Namespaces are administrative boundaries but do not provide scheduling or isolation guarantees; pods in other namespaces could still be scheduled on in-scope nodes.

Question 90

Your organization recently deployed a new application on Google Kubernetes Engine. You need to deploy a solution to protect the application. The solution has the following requirements:

• Scans must run at least once per week

• Must be able to detect cross-site scripting vulnerabilities

• Must be able to authenticate using Google accounts

Which solution should you use?

  1. Google Cloud Armor
  2. Web Security Scanner
  3. Security Health Analytics
  4. Container Threat Detection
Show answer and explanation

Correct answer: B. Web Security Scanner

Web Security Scanner is the correct solution as it meets all three requirements: it can be scheduled to run at least once per week, it detects cross-site scripting vulnerabilities along with other web application vulnerabilities, and it authenticates using Google accounts to access the application for scanning.

Why the other options are wrong

  • A. Google Cloud Armor provides DDoS protection and WAF capabilities but is not a vulnerability scanner.
  • C. Security Health Analytics scans infrastructure and configuration but does not perform web application vulnerability scanning or XSS detection.
  • D. Container Threat Detection monitors container runtime behavior but does not scan for web application vulnerabilities like XSS.

Question 91

A customer terminates an engineer and needs to make sure the engineer's Google account is automatically deprovisioned.

What should the customer do?

  1. Use the Cloud SDK with their directory service to remove their IAM permissions in Cloud Identity.
  2. Use the Cloud SDK with their directory service to provision and deprovision users from Cloud Identity.
  3. Configure Cloud Directory Sync with their directory service to provision and deprovision users from Cloud Identity.
  4. Configure Cloud Directory Sync with their directory service to remove their IAM permissions in Cloud Identity.
Show answer and explanation

Correct answer: C. Configure Cloud Directory Sync with their directory service to provision and deprovision users from Cloud Identity.

Cloud Directory Sync (GCDS) is the correct tool for automated provisioning and deprovisioning of users from Cloud Identity based on a directory service source of truth. When an employee is terminated and removed from the directory service, GCDS automatically syncs that change to Cloud Identity, deprovisioning the account. This ensures deprovisioning happens automatically without manual intervention.

Why the other options are wrong

  • A. The Cloud SDK is a command-line tool for manual operations, not for automated deprovisioning tied to directory service changes.
  • B. While the Cloud SDK can be used with a directory service, it requires manual execution and does not provide automatic deprovisioning.
  • D. Cloud Directory Sync does not remove IAM permissions; it provisions and deprovisions user accounts from Cloud Identity based on directory changes.

Question 92

Your organization hosts a financial services application running on Compute Engine instances for a third-party company. The third-party company's servers that will consume the application also run on Compute Engine in a separate Google Cloud organization. You need to configure a secure network connection between the Compute Engine instances. You have the following requirements:

• The network connection must be encrypted.

• The communication between servers must be over private IP addresses.

What should you do?

  1. Configure a Cloud VPN connection between your organization's VPC network and the third party's that is controlled by VPC firewall rules.
  2. Configure a VPC peering connection between your organization's VPC network and the third party's that is controlled by VPC firewall rules.
  3. Configure a VPC Service Controls perimeter around your Compute Engine instances, and provide access to the third party via an access level.
  4. Configure an Apigee proxy that exposes your Compute Engine-hosted application as an API, and is encrypted with TLS which allows access only to the third party.
Show answer and explanation

Correct answer: A. Configure a Cloud VPN connection between your organization's VPC network and the third party's that is controlled by VPC firewall rules.

Cloud VPN provides encrypted network connections between two VPC networks across different organizations using IPsec encryption, meeting the encryption requirement. The connection operates over private IP addresses and is controlled by VPC firewall rules, providing the secure, private connectivity needed for the financial services application between the two separate Google Cloud organizations.

Why the other options are wrong

  • B. VPC peering cannot connect networks across different organizations; it only works within the same organization or between organizations in specific limited scenarios.
  • C. VPC Service Controls provides access control and perimeter security but does not create the underlying encrypted network connection required for direct communication.
  • D. Apigee is an API management platform, not a network connection solution, and would add unnecessary complexity for direct server-to-server communication.

Question 93

You are on your company's development team. You noticed that your web application hosted in staging on GKE dynamically includes user data in web pages without first properly validating the inputted data. This could allow an attacker to execute gibberish commands and display arbitrary content in a victim user's browser in a production environment.

How should you prevent and fix this vulnerability?

  1. Use Cloud IAP based on IP address or end-user device attributes to prevent and fix the vulnerability.
  2. Set up an HTTPS load balancer, and then use Cloud Armor for the production environment to prevent the potential XSS attack.
  3. Use Web Security Scanner to validate the usage of an outdated library in the code, and then use a secured version of the included library.
  4. Use Web Security Scanner in staging to simulate an XSS injection attack, and then use a templating system that supports contextual autoescaping.
Show answer and explanation

Correct answer: D. Use Web Security Scanner in staging to simulate an XSS injection attack, and then use a templating system that supports contextual autoescaping.

Web Security Scanner simulates XSS injection attacks in the staging environment to identify the vulnerability, and contextual auto-escaping through a templating system is the proper technical fix for Cross-Site Scripting vulnerabilities. This approach validates the vulnerability exists, tests the fix, and implements a proven solution that automatically escapes user input based on context, preventing arbitrary code execution.

Why the other options are wrong

  • A. Cloud IAP controls who can access an application but does not prevent XSS attacks once authenticated users are inside the application.
  • B. Cloud Armor protects against network-layer attacks and some application attacks but is not the primary solution for XSS vulnerabilities, which require input validation and output encoding.
  • C. Web Security Scanner identifies XSS vulnerabilities but not outdated libraries; using a secured library version alone doesn't fix improper user data inclusion in web pages.

Question 94

A customer's company has multiple business units. Each business unit operates independently, and each has their own engineering group. Your team wants visibility into all projects created within the company and wants to organize their Google Cloud Platform (GCP) projects based on different business units. Each business unit also requires separate sets of IAM permissions.

Which strategy should you use to meet these needs?

  1. Create an organization node, and assign folders for each business unit.
  2. Establish standalone projects for each business unit, using gmail.com accounts.
  3. Assign GCP resources in a project, with a label identifying which business unit owns the resource.
  4. Assign GCP resources in a VPC for each business unit to separate network access.
Show answer and explanation

Correct answer: A. Create an organization node, and assign folders for each business unit.

Creating an organization node with folders for each business unit provides the hierarchical structure needed for visibility, organization, and separate IAM permissions. This approach allows the company to view all projects across the organization while maintaining business unit separation through folders. IAM permissions can be set at the folder level to ensure each business unit has independent access controls and isolation.

Why the other options are wrong

  • B. Using gmail.com accounts and standalone projects lacks the organizational structure and centralized visibility required, and is not a best practice for enterprise deployments.
  • C. Labels identify resources but don't provide organizational hierarchy or separate IAM permission boundaries like folders do.
  • D. VPC networks control network traffic; they don't organize projects or provide the hierarchical visibility and separate IAM control needed for this scenario.

That was 94 of 376.

The full Google Professional Cloud Security Engineer pack has all 376 questions, each with the answer, the explanation and why the other options are wrong, plus a questions-only copy for timed runs. US$39, paid once, with free monthly updates and a pass-or-your-money-back guarantee.

Get the full pack

Not ready yet? Get the first 180 free as one PDF by email. Get the Half Pack →