Google Cloud Certified - Professional Cloud Architect — Questions and Answers

Question 1: A company needs to connect their on-premises data center to Google Cloud with 10 Gbps bandwidth, consistent latency, and an SLA. Which option is correct?

Cloud VPN with high-availability tunnels
Partner Interconnect through a service provider
Direct Peering with Google's edge network
Dedicated Interconnect with 10 Gbps circuit (Correct answer)

Correct answer: Dedicated Interconnect with 10 Gbps circuit

Dedicated Interconnect provides a direct physical connection to Google with a 99.99% SLA (with redundant circuits) and guaranteed bandwidth without traversing the public internet.

Question 2: Which firewall rule direction handles traffic coming INTO a VM instance from outside?

Bidirectional
NAT
Ingress (Correct answer)
Egress

Correct answer: Ingress

Ingress firewall rules control inbound traffic arriving at VM instances from external sources.

Question 3: The focus of the least privilege principle is:

Ensuring that roles are defined for your company's access model
Ensuring that you give users the least amount of privilege possible
Minimizing the amount of privileges that a single user has to the bare minimum needed to perform their job duties (Correct answer)
Separating the amount of responsibilities that each of your applications has into logical chunks

Correct answer: Minimizing the amount of privileges that a single user has to the bare minimum needed to perform their job duties

Explanation: Making sure that your users have access to only what is necessary to perform their job obligations is the aim of the least privilege.

Question 4: A company needs to grant a third-party auditor read-only access to specific BigQuery datasets without giving them access to other GCP resources. What is the BEST approach?

Add the auditor to a Google Group with Editor permissions
Share the dataset using BigQuery authorized views
Grant the auditor the Project Viewer role at the project level
Create a custom IAM role with only BigQuery Data Viewer permissions and assign it at the dataset level (Correct answer)

Correct answer: Create a custom IAM role with only BigQuery Data Viewer permissions and assign it at the dataset level

Creating a custom IAM role with BigQuery Data Viewer and assigning it at the dataset level follows the principle of least privilege, scoping access precisely to what is needed.

Question 5: An architect needs to implement a blue/green deployment for a Cloud Run service to allow instant rollback. What is the correct mechanism?

Use Cloud Run traffic splitting to route 100% traffic to the new revision, retaining the old revision (Correct answer)
Use Cloud Deploy with a canary deployment pipeline
Deploy two separate Cloud Run services and update DNS manually
Use a global load balancer with two backend services

Correct answer: Use Cloud Run traffic splitting to route 100% traffic to the new revision, retaining the old revision

Cloud Run revision traffic splitting allows atomic traffic shifts between revisions, and rolling back simply redirects 100% traffic to the prior revision instantly.

Question 6: What Cloud Storage class is most cost-effective for data accessed less than once per year?

Archive (Correct answer)
Coldline
Nearline
Standard

Correct answer: Archive

Archive storage has the lowest storage cost but highest retrieval cost, making it ideal for data accessed less than once a year.

Question 7: What is the purpose of health checks in a GCP Managed Instance Group?

Automatically detect and replace unhealthy VM instances (autohealing) (Correct answer)
Monitor network throughput
Control access to VMs
Scale instances based on CPU usage

Correct answer: Automatically detect and replace unhealthy VM instances (autohealing)

MIG autohealing uses health checks to detect unhealthy instances and automatically recreate them, maintaining the desired instance count and availability.

Question 8: Which GCP monitoring concept defines a target level of service that is measured against an error budget?

KPI (Key Performance Indicator)
SLO (Service Level Objective) (Correct answer)
OKR (Objective and Key Result)
SLA (Service Level Agreement)

Correct answer: SLO (Service Level Objective)

An SLO defines the target reliability level (e.g., 99.9% availability) and is used to calculate the error budget for the service.

Question 9: What is the benefit of using regional Persistent Disks in Google Compute Engine?

Automatic snapshots included
Synchronous replication across two zones for higher durability and availability (Correct answer)
Lower cost than zonal disks
Faster I/O performance than zonal disks

Correct answer: Synchronous replication across two zones for higher durability and availability

Regional Persistent Disks replicate data synchronously across two zones in a region, providing resilience against single-zone failures.

Question 10: Which Cloud Storage feature helps protect against accidental deletion by retaining deleted objects for a configurable period?

Object Lifecycle Management
Object Versioning (Correct answer)
Retention Policies
Bucket Lock

Correct answer: Object Versioning

Object Versioning in Cloud Storage keeps a history of object versions, allowing recovery of accidentally deleted or overwritten objects.

Question 11: A team wants to run stateless containerized workloads on GCP without managing any infrastructure or Kubernetes clusters. Which service is best?

GKE Autopilot
GKE Standard
Cloud Run (Correct answer)
GCE Managed Instance Groups

Correct answer: Cloud Run

Cloud Run is a fully managed serverless platform for stateless containers that abstracts all infrastructure management.

Question 12: Which GCP feature allows you to create scheduled, automated snapshots of Persistent Disks for backup?

Snapshot Schedules (Correct answer)
Cloud Backup and DR service
Managed Instance Group autohealing
Disk cloning

Correct answer: Snapshot Schedules

Snapshot Schedules in Compute Engine automate the creation and retention of Persistent Disk snapshots on a defined schedule.

Question 13: A team wants to enforce that all Compute Engine VMs in their organization use only approved OS images. What is the correct control?

Audit images weekly with a Cloud Function script
Require manual approval for every VM creation request
Use an Organization Policy constraint to restrict allowed images (Correct answer)
Add a firewall rule blocking unapproved image traffic

Correct answer: Use an Organization Policy constraint to restrict allowed images

The constraints/compute.trustedImageProjects organization policy prevents VM creation from images outside the approved list.

Question 14: Which GCP service collects application performance data such as latency and request traces across distributed services?

Cloud Logging
Error Reporting
Cloud Monitoring
Cloud Trace (Correct answer)

Correct answer: Cloud Trace

Cloud Trace collects distributed tracing data to help analyze latency across microservices and identify performance bottlenecks.

Question 15: A financial company needs an RTO of less than 15 minutes for their GCP-hosted trading platform. Which architecture best achieves this?

Hot standby in a second region with automated health-check-driven failover (Correct answer)
Daily snapshots with 2-hour restore procedures
Cross-region read replica promoted manually
Warm standby with manual failover scripts

Correct answer: Hot standby in a second region with automated health-check-driven failover

A hot standby with automated failover enables sub-minute failover detection and traffic rerouting, well within a 15-minute RTO.

Question 16: Your organization requires that only users from specific geographic regions can access a public-facing GCP application. Which service enforces this?

Cloud IAP
Cloud Armor geo-restriction policies (Correct answer)
VPC Firewall Rules
Cloud DNS

Correct answer: Cloud Armor geo-restriction policies

Cloud Armor supports geo-based access control policies that allow or deny traffic based on the request's origin country.

Question 17: A financial application requires full ACID compliance and the ability to scale to millions of transactions per second globally. Which database is the best fit?

BigQuery
Cloud Spanner (Correct answer)
Firestore
Cloud SQL

Correct answer: Cloud Spanner

Cloud Spanner provides global distribution with full ACID transactions and can handle millions of TPS, making it ideal for mission-critical financial apps.

Question 18: What is a key characteristic of a Warm Standby disaster recovery strategy?

Backups are replicated but no compute runs in the DR region
The environment is created only after a disaster is declared
The standby environment is identical to production and always running at full capacity
A scaled-down version of the environment runs continuously and can scale up quickly (Correct answer)

Correct answer: A scaled-down version of the environment runs continuously and can scale up quickly

Warm Standby keeps a reduced-capacity version of the environment running, allowing faster recovery than a cold standby but at lower cost than hot standby.

Question 19: You examined TerramEarth's need for downtime reduction and discovered that they can save the majority of the time by cutting down on client wait times for parts. You made the decision to concentrate on cutting the total reporting time for the three weeks. Which process changes for the company should you suggest?

80% more fleet cellular connectivity, a move from FTP to streaming transport, and the development of metrics machine learning analysis. (Correct answer)
Convert from FTP to SFTP transport, create metrics analysis using machine learning, and boost dealer local inventory by a fixed percentage.
Convert measurements to machine learning for analysis, switch from FTP to SFTP transport, and migrate from CSV to binary format.
Convert CSV to binary format, move from FTP to streaming transport, and create metrics machine learning analysis.

Correct answer: 80% more fleet cellular connectivity, a move from FTP to streaming transport, and the development of metrics machine learning analysis.

Explanation: Since the data is now only collected when the machines are down for maintenance, having cellular connectivity will significantly increase how recent the data is when it is utilized for analysis. Using streaming transport as opposed to recurrent FTP will further tighten the feedback loop. Workloads requiring predictive maintenance are perfect for machine learning.

Question 20: Which GKE feature helps ensure a minimum number of pods remain available during voluntary disruptions like node upgrades?

Horizontal Pod Autoscaler
Resource Quotas
Pod Disruption Budget (PDB) (Correct answer)
LimitRange

Correct answer: Pod Disruption Budget (PDB)

Pod Disruption Budgets define the minimum number of pods that must remain available during voluntary disruptions, preventing outages during maintenance.

Question 21: A Kubernetes platform solution from IBM Red Hat called OpenShift is going to be implemented by your organization across all of its on-premises and public cloud environments. What kind of shared responsibility model will this deployment entail for you, as the lead architect accountable for your company's GCP deployments?

PaaS
On premises
SaaS
IaaS (Correct answer)

Correct answer: IaaS

Explanation: The important thing to keep in mind is that in order for a service provider (in this case, GCP) to assume responsibility for its PaaS, the service must be made available as a managed service. GKE, a Kubernetes platform provided by GCP, is available. OpenShift, however, is not a PaaS service provided by Google. As a result, Google will not be held accountable for the architecture and internal workings of your OpenShift setups. All of the virtual machines that OpenShift will deploy as part of its GCP deployment must be managed. In light of the shared responsibility approach, this implementation is an IaaS.

Question 22: What term best sums up a Google Cloud Architect's job?

A technical subject matter expert that knows business and technological requirements and is able to translate those requirements into technical designs
A person with the ability to create, manage, and use strong, secure, scalable, highly available, and dynamic solutions to advance business goals (Correct answer)
An engineer who is interested in designing robust solutions in an enterprise
An individual within an organization who architects

Correct answer: A person with the ability to create, manage, and use strong, secure, scalable, highly available, and dynamic solutions to advance business goals

Explanation: Although all answers are technically valid, the question is seeking the best definition of the work of a Google Cloud Architect, which is an individual who can design, construct and manage resilient, secure, scalable, highly available, and dynamic solutions to drive business objectives.

Question 23: What is the function of Cloud Router in Google Cloud?

Routes DNS queries
Dynamically exchanges routes between VPC and on-premises networks using BGP (Correct answer)
Monitors network traffic
Manages firewall rules

Correct answer: Dynamically exchanges routes between VPC and on-premises networks using BGP

Cloud Router uses Border Gateway Protocol (BGP) to dynamically advertise and learn routes between Google Cloud and on-premises networks.

Question 24: An architect needs to ensure that a multi-project GCP environment sends all security findings to a single Security Operations Center. Which service aggregates findings?

Security Command Center (SCC) (Correct answer)
VPC Service Controls
Cloud Monitoring workspace
Cloud Logging sink to BigQuery

Correct answer: Security Command Center (SCC)

Security Command Center (SCC) provides a centralized view of security and compliance findings across all GCP projects in an organization.

Question 25: Which feature of GKE allows automatic node pool creation based on pending pod requirements?

Horizontal Pod Autoscaler
Vertical Pod Autoscaler
Managed Instance Group autoscaling
Node Auto-Provisioner (NAP) (Correct answer)

Correct answer: Node Auto-Provisioner (NAP)

Node Auto-Provisioner (NAP) is a feature of GKE Cluster Autoscaler that automatically creates optimally sized node pools for unschedulable pods.

Question 26: Your GKE workload needs to access a Cloud SQL instance securely without exposing the database to the public internet. What is the recommended approach?

Assign a public IP to Cloud SQL and whitelist node IP ranges
Use the Cloud SQL Auth Proxy as a sidecar container with Workload Identity (Correct answer)
Use Cloud SQL with a private IP and connect via the internal VPC
Open a VPN tunnel from each pod to the Cloud SQL instance

Correct answer: Use the Cloud SQL Auth Proxy as a sidecar container with Workload Identity

The Cloud SQL Auth Proxy sidecar handles IAM authentication and TLS, while Workload Identity eliminates the need to manage service account key files.

Question 27: A company wants to run a batch data processing job on GCP that needs to start quickly and complete within a few hours at lowest cost. Which VM option is most appropriate?

Committed Use Discounts
Standard VMs with on-demand pricing
Preemptible VMs (Spot VMs) (Correct answer)
Sole-tenant nodes

Correct answer: Preemptible VMs (Spot VMs)

Preemptible (Spot) VMs offer significant discounts (up to 91%) for short-lived workloads that can tolerate interruption, ideal for batch processing.

Question 28: A GKE workload requires access to a Cloud Storage bucket. What is the recommended way to grant this permission without using service account keys?

Use Workload Identity to bind a Kubernetes service account to a GCP service account (Correct answer)
Grant roles to the default compute service account
Mount a service account key file as a Kubernetes secret
Use Cloud IAP

Correct answer: Use Workload Identity to bind a Kubernetes service account to a GCP service account

Workload Identity is the recommended approach for GKE pods to authenticate as a GCP service account without managing key files.

Question 29: You are creating a conversation app for mobile devices. By demonstrating that a message was delivered by a certain user, you may prevent chat messages from being spoofed. What ought you to do?

Tag messages client-side with the originating user identifier and the destination user.
Use a trusted certificate authority to enable SSL connectivity between the client application and the server.
Use public key infrastructure (PKI) to encrypt the message client-side using the originating user’s private key. (Correct answer)
Encrypt the message client-side using block-based encryption with a shared key.

Correct answer: Use public key infrastructure (PKI) to encrypt the message client-side using the originating user’s private key.

Explanation: PKI mandates that the server and client both possess signed certificates that validate both the server and the client.

Question 30: Which Cloud SQL feature provides automatic failover to a standby replica in case of a primary instance failure?

Cloud SQL Backups
Cross-region Read Replica
Point-in-time recovery
High Availability (HA) with failover replica (Correct answer)

Correct answer: High Availability (HA) with failover replica

Cloud SQL High Availability creates a synchronous standby replica in the same region that automatically takes over if the primary fails.

Question 31: Which GCP service is optimized for petabyte-scale, low-latency reads/writes for time-series or wide-column workloads like IoT data?

Cloud Spanner
BigQuery
Cloud Bigtable (Correct answer)
Cloud SQL

Correct answer: Cloud Bigtable

Cloud Bigtable is a managed wide-column NoSQL database designed for petabyte-scale time-series and analytical workloads with low latency.

Question 32: Which GKE feature automatically adjusts pod CPU and memory requests based on historical usage?

Node Auto-Provisioner
Vertical Pod Autoscaler (Correct answer)
Horizontal Pod Autoscaler
Cluster Autoscaler

Correct answer: Vertical Pod Autoscaler

Vertical Pod Autoscaler (VPA) recommends and optionally applies CPU and memory request adjustments based on actual pod usage history.

Question 33: Which GCP multi-region option for Cloud Storage ensures data is stored redundantly across at least two geographic regions?

Regional storage class
Dual-region or Multi-region location (Correct answer)
Nearline cross-region
Coldline multi-AZ

Correct answer: Dual-region or Multi-region location

Dual-region and multi-region Cloud Storage locations replicate data across geographically separated regions for highest durability and availability.

Question 34: A healthcare application must ensure that data never leaves a specific geographic region for compliance. Which combination of controls enforces this?

Deploy only in that region and trust developers to comply
Use VPC Service Controls with a perimeter and resource location restrictions via Organization Policy (Correct answer)
Enable Cloud Armor geo-blocking for API requests
Label all resources with the region name

Correct answer: Use VPC Service Controls with a perimeter and resource location restrictions via Organization Policy

VPC Service Controls restrict API access to within the perimeter, and the constraints/gcp.resourceLocations policy prevents resource creation outside approved regions.

Question 35: Which IAM binding type allows you to set conditions on when a role applies to a member, such as restricting access to specific hours or IP ranges?

Policy bindings
Conditional bindings (Correct answer)
Service account bindings
Basic bindings

Correct answer: Conditional bindings

Conditional bindings (IAM Conditions) allow you to define attribute-based access control rules that restrict when a role applies based on request attributes like time, IP, or resource tags.

Question 36: What is the purpose of Log-Based Metrics in Cloud Logging?

Store logs in BigQuery automatically
Filter logs by severity level
Send logs to Pub/Sub topics
Create Cloud Monitoring metrics from log entry data (Correct answer)

Correct answer: Create Cloud Monitoring metrics from log entry data

Log-Based Metrics extract metric data from log entries in Cloud Logging and make them available in Cloud Monitoring for alerting and dashboards.

Question 37: A GCP architect needs to give on-premises users low-latency access to a Cloud Storage bucket without traversing the internet. What should they use?

Cloud CDN
Cloud DNS
Cloud VPN only
Direct Peering or Cloud Interconnect with Private Google Access for on-premises (Correct answer)

Correct answer: Direct Peering or Cloud Interconnect with Private Google Access for on-premises

Cloud Interconnect combined with Private Google Access for on-premises enables low-latency, private access to Google services like Cloud Storage.

Question 38: A team wants to debug a production Cloud Run service without redeploying or adding log statements. Which GCP tool allows setting breakpoints and inspecting state?

Cloud Debugger (Snapshot Debugger) (Correct answer)
Error Reporting
Cloud Profiler
Cloud Trace

Correct answer: Cloud Debugger (Snapshot Debugger)

Cloud Debugger (now Snapshot Debugger) allows inspecting application state at specific code locations in running apps without stopping or redeploying.

Question 39: A startup needs a fully managed, serverless NoSQL document database for a mobile app with real-time sync. Which service fits best?

Memorystore
Firestore (Correct answer)
Cloud Spanner
Cloud Bigtable

Correct answer: Firestore

Firestore is a serverless, NoSQL document database with real-time synchronization, ideal for mobile and web applications.

Question 40: A company has an RPO of 4 hours for a Cloud SQL database. What backup strategy satisfies this requirement at minimum cost?

Daily backups with manual restore procedures
Automated backups every 4 hours with point-in-time recovery enabled (Correct answer)
Continuous replication to a read replica
Synchronous multi-region replication

Correct answer: Automated backups every 4 hours with point-in-time recovery enabled

Automated backups every 4 hours combined with point-in-time recovery ensures data loss is limited to at most 4 hours, meeting the RPO requirement.

Question 41: Which GCP service provides a managed Redis or Memcached in-memory cache to reduce database load?

Firestore
Cloud SQL
Cloud Bigtable
Memorystore (Correct answer)

Correct answer: Memorystore

Memorystore is Google Cloud's fully managed Redis and Memcached service for low-latency in-memory caching.

Question 42: Your company wants to introduce a number of microservices to help its system handle varying demands. A distinct version of software libraries is used by each microservice. You want to make it possible for their developers to maintain compatibility between their development environment and the numerous production services. What kind of technology should you pick?

RPM/DEB
Containers (Correct answer)
Virtual machines
Chef/Puppet

Correct answer: Containers

Explanation: By abstracting away system OS environments with containers, development, test, and production deployments can all use the same host OS image. A copy-on-write filesystem is used to record changes made during development, and teams may easily publish updated versions of the microservices in a repository.

Question 43: What is the Recovery Time Objective (RTO) in the context of disaster recovery planning?

The time interval between backups
The maximum acceptable time to restore service after a disaster (Correct answer)
The maximum acceptable amount of data loss measured in time
The minimum number of replicas required for redundancy

Correct answer: The maximum acceptable time to restore service after a disaster

RTO defines the maximum tolerable downtime — how quickly the service must be restored after a disaster occurs.

Question 44: Which AlloyDB feature makes it significantly faster than standard Cloud SQL PostgreSQL for analytical queries?

Columnar engine for HTAP workloads (Correct answer)
Built-in Redis cache
Multi-region synchronous writes
Automatic sharding

Correct answer: Columnar engine for HTAP workloads

AlloyDB's columnar engine accelerates analytical queries by storing data in a column-oriented format in memory alongside the row store.

Question 45: A Kubernetes deployment on GKE needs to ensure pods are spread across multiple zones for high availability. Which feature achieves this?

Node Affinity rules
Cluster Autoscaler
Topology Spread Constraints (Correct answer)
Pod Disruption Budgets

Correct answer: Topology Spread Constraints

Topology Spread Constraints control how pods are distributed across topology domains like zones, ensuring high availability.

Question 46: A company needs to export all GCP logs to a centralized BigQuery dataset for long-term analysis. What is the best approach?

Use Cloud Pub/Sub to push logs to BigQuery
Configure a Log Sink to route logs to BigQuery (Correct answer)
Download logs manually via the console
Use Cloud Dataflow to read logs from Cloud Logging

Correct answer: Configure a Log Sink to route logs to BigQuery

Log Sinks in Cloud Logging automatically route log entries to destinations like BigQuery, Cloud Storage, or Pub/Sub for archival and analysis.

Question 47: On Google Cloud Platform, your company has a 3-tier web application installed on the same network. The web, API, and database tiers can scale separately from one another. The web should be the conduit for network traffic that travels to the API tier and ultimately to the database tier. The database and web tiers shouldn't be in communication with one another. How should the network be configured?

Set up software-based firewalls on individual VMs
Add tags to each tier and set up routes to allow the desired traffic flow
Add tags to each tier and set up firewall rules to allow the desired traffic flow (Correct answer)
Add each tier to a different subnetwork.

Correct answer: Add tags to each tier and set up firewall rules to allow the desired traffic flow

Explanation: Each instance will have the same tag to identify the tier as it grows. Since tags can be used for both the target and the source, firewall rules can make use of them to allow or restrict traffic as needed.

Question 48: Which Google Cloud feature allows you to share a VPC network across multiple projects within the same organization?

Shared VPC (Correct answer)
Cloud NAT
Cloud Interconnect
VPC Peering

Correct answer: Shared VPC

Shared VPC lets an organization connect resources from multiple projects to a common VPC network, enabling centralized network administration.

Question 49: What is the purpose of a Cloud Monitoring workspace?

Manage billing for monitored resources
Define IAM roles for Logging access
Provide a unified monitoring view across multiple GCP projects (Correct answer)
Store audit log archives

Correct answer: Provide a unified monitoring view across multiple GCP projects

A Cloud Monitoring workspace aggregates metrics and dashboards from multiple GCP projects into a single monitoring interface.

Question 50: Which type of Cloud VPN is recommended for high-availability production workloads?

Classic VPN
Dedicated VPN
Direct VPN
HA VPN (Correct answer)

Correct answer: HA VPN

HA VPN provides a 99.99% SLA by using two VPN gateways with redundant tunnels, making it suitable for production environments.

Question 51: Which Cloud Run feature allows you to deploy Cloud Run services within a VPC for private access to internal resources?

Cloud Run Private Endpoints
Cloud Run Shared VPC
Cloud Run Direct VPC Egress (Correct answer)
VPC Connector (Serverless VPC Access)

Correct answer: Cloud Run Direct VPC Egress

Cloud Run Direct VPC Egress allows Cloud Run services to send traffic directly into a VPC without needing a Serverless VPC Access connector.

Question 52: A startup needs an event-driven architecture where multiple downstream services react to new user registrations. What is the recommended pattern on Google Cloud?

Send HTTP webhooks from the registration service to each consumer
Use Cloud Scheduler to trigger each service on a fixed interval
Write events directly to a shared Cloud SQL table that each service polls
Publish to Pub/Sub topic with each service having its own subscription (Correct answer)

Correct answer: Publish to Pub/Sub topic with each service having its own subscription

Pub/Sub fan-out with per-service subscriptions decouples producers from consumers and guarantees at-least-once delivery to each subscriber independently.

Google Cloud Certified - Professional Cloud Architect

The Professional Cloud Architect certification validates the ability to design, develop, and manage robust, secure, scalable, and cost-effective cloud solutions on Google Cloud. It tests knowledge of cloud architecture, infrastructure management, security, compliance, and the Google Cloud Well-Architected Framework.

Questions

120 min

Time Limit

70%

To Pass

Checking for saved progress...

Exam Rules

You can skip questions and return to them later
Flag questions for review before submitting
No feedback shown until you submit the entire exam
Unanswered questions count as wrong — answer everything
10 pretest questions are mixed in and don't affect your score
Timer auto-submits when time runs out
Your progress is auto-saved every 30 seconds

These practice questions were written with AI tools. They are not real exam questions. Exam facts are checked against the official sources listed below. Not reviewed by a subject-matter expert. Editorial policy

Sources & references

Exam facts on this page (fees, format, dates, passing scores) are checked against these official sources; the date is our last check. Editorial policy

pearsonvue.com — Exam information · verified Nov 2026
support.google.com — Passing score, score validity, retake fee, exam information · verified Nov 2026
cloud.google.com — Test delivery, eligibility, fees, exam format, time limit, score validity · verified Nov 2026