Sign in
Home/Google Cloud/Professional Cloud Architect (Google Cloud)/Free questions

Professional Cloud Architect (Google Cloud) — Free Practice Questions

10 free sample questions from a bank of 279, with the correct answers and explanations. No signup required — start practising right now.

1Your company has decided to make a major revision of their API in order to create better experiences for their developers. They need to keep the old version of the API available and deployable, while allowing new customers and testers to try out the new API. They want to keep the same SSL and DNS records in place to serve both APIs. What should they do?
  • Configure a new load balancer for the new version of the API
  • Reconfigure old clients to use a new endpoint for the new API
  • Have the old API forward traffic to the new API based on the path
  • Use separate backend pools for each API path behind the load balancer
Answer: D

The short version

D — Separate backend pools per API path behind one LB. Same SSL and DNS stay while paths route to old versus new backends for parallel serving.

Key concepts in this question

  • Single endpoint constraint: keep SSL cert and DNS stable while exposing two API versions
  • Path-based routing: load balancer forwards by URL path to distinct backend services
  • Backend pools: isolate versions for independent deploy and scale without client changes
  • Forwarding or new LBs: either chains versions or fragments the endpoint, breaking the requirement

Why D is correct

Keeping the same DNS and cert while serving both APIs demands Layer 7 split on one frontend. Separate backends per path let old and new scale and roll independently while testers use new paths. That is the standard PCA pattern for API revision coexistence. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • A. adds a second frontend, splitting DNS and cert management
  • B. forces old clients to reconfigure, violating keep-old-available transparently
  • C. couples new to old availability and complicates rollback versus independent pools

PCA exam tip

Same DNS plus two API versions equals one LB with path backends: split by path, keep cert, deploy independently.

2Your company plans to migrate a multi-petabyte data set to the cloud. The data set must be available 24hrs a day. Your business analysts have experience only with using a SQL interface. How should you store the data to optimize it for ease of analysis?
  • Load data into Google BigQuery
  • Insert data into Google Cloud SQL
  • Put flat files into Google Cloud Storage
  • Stream data into Google Cloud Datastore
Answer: A

The short version

A — Load multi-petabyte SQL data into BigQuery. Serverless analytics with SQL interface handles always-on massive sets without managing clusters.

Key concepts in this question

  • Analyst skill: SQL-only team needs familiar querying without new APIs
  • Scale and availability: petabytes 24 hours demand distributed serverless storage and compute
  • BigQuery: columnar warehouse with SQL, scaling to petabytes and continuous serving
  • SQL, files, and Datastore: limit scale, SQL richness, or operational simplicity for this size

Why A is correct

BigQuery stores massive data durably and queries it with standard SQL the analysts already know. It stays available without cluster care and separates storage from compute for cost control. Flat files or provisioned SQL cannot match that blend of scale, SQL ease, and always-on access. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • B. caps scale and ops for petabyte analytics versus serverless warehouse
  • C. leaves SQL access and performance tuning to the team on raw objects
  • D. targets operational key-value lookups, not large-scale SQL analysis

PCA exam tip

Petabytes plus SQL-only plus always-on equals BigQuery: serverless SQL warehouse for massive analytics.

3The operations manager asks you for a list of recommended practices that she should consider when migrating a J2EE application to the cloud. Which three practices should you recommend? (Choose three.)
  • Port the application code to run on Google App Engine
  • Integrate Cloud Dataflow into the application to capture real-time metrics
  • Instrument the application with a monitoring tool like Stackdriver Debugger
  • Select an automation framework to reliably provision the cloud infrastructure
  • Deploy a continuous integration tool with automated testing in a staging environment
  • Migrate from MySQL to a managed NoSQL database like Google Cloud Datastore or Bigtable
Answer: C, D, E

The short version

CDE — Instrument, automate infra, and add CI with staging. Observability plus IaC plus tested pipelines de-risks J2EE moves to cloud.

Key concepts in this question

  • Monitoring instrumentation: Stackdriver Debugger and tracing expose cloud behavior early
  • Automation framework: reliably provisions networks and VMs repeatably instead of click-ops
  • CI with staging and tests: catches regressions before production on every change
  • Rewrite or DB swap: porting to App Engine or changing to NoSQL adds risk without asked benefit

Why CDE is correct

Migrating J2EE safely hinges on seeing what happens, recreating infra identically, and validating continuously. Instrumentation, automation, and CI with staging together cover those pillars. Dataflow metrics or forced NoSQL shifts distract from core migration hygiene. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • A. rewrites for a specific PaaS, unnecessary for lift-and-shift best practice
  • B. adds streaming metrics plumbing not central to migration safety
  • F. changes data paradigm without requirement, raising conversion risk

PCA exam tip

J2EE to cloud best trio equals monitor, automate, and CI-test: observe, provision as code, validate in staging.

4A news feed web service has the following code running on Google App Engine. During peak load, users report that they can see news articles they already viewed. What is the most likely cause of this problem?
  • The session variable is local to just a single instance
  • The session variable is being overwritten in Cloud Datastore
  • The URL of the API needs to be modified to prevent caching
  • The HTTP Expires header needs to be set to -1 stop caching
Answer: A

The short version

A — Local session state causes repeat views at peak. Each App Engine instance keeps its own session, so users hitting different instances see stale progress.

Key concepts in this question

  • Instance-local variables: memory tied to one VM, invisible to other scaled-out instances
  • Peak load scaling: many instances serve traffic, scattering a user across them
  • Shared session need: viewed-article state must live in Memcache, Datastore, or cookies
  • Caching headers: affect browser and CDN freshness, not per-instance session divergence

Why A is correct

Seeing already-viewed articles only under peak points to scale-out splitting sessions. When the viewed set lives locally, a request routed to a fresh instance lacks history and resends old items. Centralizing session fixes it, confirming local state as the likely cause. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • B. misplaces session in Datastore overwrite rather than locality across instances
  • C. URL caching tweaks do not explain instance-dependent viewed state
  • D. Expires header controls browser cache, not server-side session spread

PCA exam tip

Peak-only stale session equals local instance memory: share sessions centrally when App Engine scales out.

5An application development team believes their current logging tool will not meet their needs for their new cloud-based product. They want a better tool to capture errors and help them analyze their historical log data. You want to help them find a solution that meets their needs. What should you do?
  • Direct them to download and install the Google StackDriver logging agent
  • Send them a list of online resources about logging best practices
  • Help them define their requirements and assess viable logging tools
  • Help them upgrade their current tool to take advantage of any new features
Answer: C

The short version

C — Help define needs then assess logging tools. Requirements-first selection avoids buying features that miss error and history goals.

Key concepts in this question

  • Requirements definition: retention, query, alerting, and analysis needs for cloud errors
  • Tool assessment: compares Stackdriver, third party, and current upgrades against those needs
  • Prescriptive install: pushes a tool before knowing gaps, risking repeat failure
  • Resource lists and blind upgrades: educate or refresh but do not ensure fit

Why C is correct

The team believes the current tool will not meet new cloud demands, so the architect should structure what better means. Defining capture, history, and analysis criteria then evaluating viable options yields a defensible choice. Directing to an agent or links skips that fit analysis. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • A. installs an agent that may repeat the same capability gaps
  • B. shares generic reading without driving a selection decision
  • D. upgrades a tool already judged insufficient for new needs

PCA exam tip

Team says logging insufficient equals define requirements then assess: needs before product, never install blind.

6You need to reduce the number of unplanned rollbacks of erroneous production deployments in your company's web hosting platform. Improvement to the QA/ Test processes accomplished an 80% reduction. Which additional two approaches can you take to further reduce the rollbacks? (Choose two.)
  • Introduce a green-blue deployment model
  • Replace the QA environment with canary releases
  • Fragment the monolithic platform into microservices
  • Reduce the platformג€™s dependency on relational database systems
  • Replace the platformג€™s relational database systems with a NoSQL database
Answer: A, C

The short version

AC — Add green-blue plus decompose to microservices. Safer releases plus smaller blast radius further cut rollbacks after QA gains.

Key concepts in this question

  • Green-blue deployments: parallel prod stacks allow instant switch and fast rollback on error
  • Microservice split: isolates faults so one bad service does not roll back the whole monolith
  • Canary versus QA replacement: canaries complement but replacing QA removes valuable testing
  • DB swaps: NoSQL or fewer relational ties do not directly prevent bad deploys

Why AC is correct

QA already cut 80 percent, leaving deployment safety and scope as next levers. Green-blue makes releases reversible in seconds while microservices shrink each release impact. Together they systematically reduce unplanned full rollbacks beyond testing alone. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • B. discards QA environment, losing pre-prod validation that achieved prior gains
  • D. trims DB dependencies without fixing release mechanics
  • E. migrates data tech without addressing deployment risk

PCA exam tip

Fewer rollback deploys after QA wins equals green-blue plus microservices: safer switch plus smaller blast radius.

7To reduce costs, the Director of Engineering has required all developers to move their development infrastructure resources from on-premises virtual machines (VMs) to Google Cloud Platform. These resources go through multiple start/stop events during the day and require state to persist. You have been asked to design the process of running a development environment in Google Cloud while providing cost visibility to the finance department. Which two steps should you take? (Choose two.)
  • Use the - -no-auto-delete flag on all persistent disks and stop the VM
  • Use the - -auto-delete flag on all persistent disks and terminate the VM
  • Apply VM CPU utilization label and include it in the BigQuery billing export
  • Use Google BigQuery billing export and labels to associate cost to groups
  • Store all state into local SSD, snapshot the persistent disks, and terminate the VM
  • Store all state in Google Cloud Storage, snapshot the persistent disks, and terminate the VM
Answer: A, D

The short version

AD — Preserve disks on stop and track cost with labels plus export. No-auto-delete keeps state through start-stop while BigQuery billing shows finance per team.

Key concepts in this question

  • Persistent disks: retain dev state when VMs stop if auto-delete is disabled
  • Start-stop pattern: multiple daily cycles demand cheap pause without data loss
  • Labels plus billing export: attribute BigQuery-exported costs to groups for visibility
  • Auto-delete and local SSD: discard state or complicate persistence, raising rebuild cost

Why AD is correct

Dev VMs need state across frequent stops without paying while stopped. Keeping disks on stop preserves work while stopping the VM halts compute billing. Adding labels fed to BigQuery export gives finance per-group cost views. That pair meets persistence plus visibility goals.

Why the others are wrong

  • B. deletes disks on terminate, losing required state daily
  • C. labels CPU utilization rather than cost-ownership groups for billing
  • E. relies on ephemeral local SSD plus snapshots, slower and riskier than persistent disks
  • F. moves state to object storage with snapshot churn, overengineering simple dev persistence

PCA exam tip

Dev start-stop with state plus cost view equals keep disks on stop plus labeled billing export.

8Your company wants to track whether someone is present in a meeting room reserved for a scheduled meeting. There are 1000 meeting rooms across 5 offices on 3 continents. Each room is equipped with a motion sensor that reports its status every second. The data from the motion detector includes only a sensor ID and several different discrete items of information. Analysts will use this data, together with information about account owners and office locations. Which database type should you use?
  • Flat file
  • NoSQL
  • Relational
  • Blobstore
Answer: B

The short version

B — Use NoSQL for high-rate sensor plus join data. Massive small writes with later joins fit schemaless scale and flexible analytics.

Key concepts in this question

  • Sensor velocity: 1000 rooms every second creates sustained write throughput needing horizontal scale
  • Discrete plus relational context: sensor IDs join to owners and offices for analysis
  • NoSQL scale: handles high ingest and varied attributes without rigid schema locks
  • Flat, relational, blob: either lack query power, scale, or structure for this hybrid use

Why B is correct

Tracking presence from per-second motion events across continents is a high-write, semi-structured workload with analytical joins. NoSQL stores raw events scalably while supporting lookups joined to account and location data. Relational would strain on ingest while flat files and blobs hinder querying. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • A. cannot query and join thousands of events per second effectively
  • C. enforces schema and scale limits costly for 1000-per-second sensor streams
  • D. stores objects without indexed querying for presence analytics

PCA exam tip

Per-second sensors plus joins equals NoSQL: ingest at scale, keep IDs flexible, join context for analysts.

9You set up an autoscaling instance group to serve web traffic for an upcoming launch. After configuring the instance group as a backend service to an HTTP(S) load balancer, you notice that virtual machine (VM) instances are being terminated and re-launched every minute. The instances do not have a public IP address. You have verified the appropriate web response is coming from each instance using the curl command. You want to ensure the backend is configured correctly. What should you do?
  • Ensure that a firewall rules exists to allow source traffic on HTTP/HTTPS to reach the load balancer.
  • Assign a public IP to each instance and configure a firewall rule to allow the load balancer to reach the instance public IP.
  • Ensure that a firewall rule exists to allow load balancer health checks to reach the instances in the instance group.
  • Create a tag on each instance with the name of the load balancer. Configure a firewall rule with the name of the load balancer as the source and the instance tag as the destination.
Answer: C

The short version

C — Allow LB health checks through the firewall. Private instances churn when checks cannot reach them, so open the checker range to the group.

Key concepts in this question

  • Health checks: LB probes that mark backends healthy only on successful response
  • Private backends: no public IP means only explicitly allowed LB ranges can probe
  • Curl success locally: proves app works, isolating the issue to check path versus app
  • Frontend rules: allowing traffic to the LB does not help checks to instances

Why C is correct

Per-minute terminate and relaunch with good curl output signals failed health checks, not app failure. Without a rule permitting Google check ranges to the instance group, every VM looks unhealthy and gets replaced. Adding that firewall path stabilizes the backend. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • A. opens the frontend further while instance checks remain blocked
  • B. adds public IPs unnecessarily instead of allowing private check ranges
  • D. misuses tags and LB names as sources rather than check IP ranges

PCA exam tip

Healthy curl but LB recycles private VMs equals firewall for health checks: allow probe ranges to backends.

10You write a Python script to connect to Google BigQuery from a Google Compute Engine virtual machine. The script is printing errors that it cannot connect to BigQuery. What should you do to fix the script?
  • Install the latest BigQuery API client library for Python
  • Run your script on a new virtual machine with the BigQuery access scope enabled
  • Create a new service account with BigQuery access and execute your script with that user
  • Install the bq component for gcloud with the command gcloud components install bq.
Answer: C

The short version

C — Run with a BigQuery-enabled service account. Compute Engine scopes plus IAM decide BigQuery access, not libraries or bq components.

Key concepts in this question

  • Service accounts: identities that carry IAM roles for API calls from VMs
  • BigQuery access: requires bigquery permissions on the calling identity plus network reach
  • Scopes: VM access scopes must allow cloud API calls alongside IAM
  • Client libs and bq CLI: help coding and admin but do not grant permissions

Why C is correct

Cannot connect to BigQuery from GCE with script errors points to identity lacking data roles. Creating and using a service account with BigQuery access supplies authenticated authorized calls. Reinstalling libraries or adding CLI components leaves the permission gap unchanged. This reflects Google Cloud architecture principles to use serverless, path-based, and policy-driven design for scale and cost control.

Why the others are wrong

  • A. refreshes code packaging without fixing IAM authorization
  • B. moves VMs for scopes alone while still missing the right service-account roles
  • D. installs CLI tooling that does not confer API permissions

PCA exam tip

GCE to BigQuery denied equals service-account IAM: grant BigQuery roles to the runtime identity first.

Want the full bank of 279 questions for Professional Cloud Architect (Google Cloud)? See all practice exams.