Courseiva

Google Professional Cloud Developer (PCD) — Questions 526–600

964 questions total · 13pages · All types, answers revealed

Take a mock examExam hub
Page 7

Page 8 of 13

Page 9
526
Matchingmedium

Match each Cloud CDN feature to its benefit.

Drag a concept onto its matching description — or click a concept then click the description.

Concepts
Matches

Remove outdated content from edge caches

Authorize temporary access to private content

Serve content from any HTTP(S) server

Define how to cache different variations of content

Store content closer to users for low latency

Why these pairings

Cloud CDN features like caching, SSL/TLS termination, and compression each provide distinct benefits. Caching reduces latency, SSL/TLS termination improves security, and compression reduces bandwidth. Distractors often mix up these benefits.

Full explanation →
527
Multi-Selecthard

A company runs a stateful application on Compute Engine. They need to achieve an RPO of less than 15 minutes and an RTO of less than 30 minutes for a regional disaster. Which three steps should they include in their disaster recovery plan? (Select exactly 3.)

Select 3 answers
A.Use a managed instance group in multiple zones within the same region
B.Develop custom scripts to replicate application data asynchronously to another region
C.Configure persistent disk snapshots to a different region
D.Use regional persistent disks to replicate data within the region
E.Configure Cloud DNS with geo-routing to direct traffic to a healthy region
AnswersB, C, E

Asynchronous replication to another region can meet RPO and allow failover to that region.

Why this answer

Asynchronous replication of application data to another region can achieve an RPO of less than 15 minutes and an RTO of less than 30 minutes, as it allows the application to fail over to a secondary region with minimal data loss. Custom scripts can control replication frequency and ensure data consistency, meeting the strict RPO requirement.

Exam trap

The PCD exam often tests the distinction between regional and multi-region disaster recovery, where candidates mistakenly choose intra-region solutions like regional persistent disks or multi-zone instance groups for a regional disaster scenario.

Full explanation →
528
MCQmedium

A company uses Cloud Spanner for global transaction processing. They need to export a full database backup to GCS in a portable format for long-term archival. Which method should they use?

A.Use gcloud beta spanner databases export command to export to GCS in CSV format.
B.Use Dataflow to read the Spanner database and write to GCS in JSON format.
C.Use Cloud Spanner backups (built-in) which export to GCS in Avro format.
D.Use Cloud SQL export to GCS.
AnswerC

Cloud Spanner's managed backup feature exports the entire database to GCS in Avro files, which is portable.

Full explanation →
529
MCQmedium

A company is running a global application on Cloud Spanner. They notice high write latency on a specific table because a frequently updated row is being accessed by many clients simultaneously. Which design pattern should they implement to distribute writes across multiple nodes and reduce contention?

A.Increase the number of nodes in the Cloud Spanner instance.
B.Use interleaved tables to co-locate related data.
C.Add a hash suffix to the primary key of the hot row to split it into multiple rows.
D.Migrate the table to Cloud Bigtable which handles hotspots better.
AnswerC

This distributes writes across multiple splits.

Why this answer

Adding a hash suffix to the primary key of the hot row splits the single heavily contended row into multiple logical rows, each with a different primary key. This distributes the write load across multiple Cloud Spanner splits and nodes, reducing lock contention and write latency. Cloud Spanner uses a distributed, synchronous replication architecture where a single row is managed by a single split; splitting the hot row into multiple rows allows parallel writes to different splits.

Exam trap

The PCD exam often tests the misconception that scaling up the instance (adding nodes) solves single-row contention, but the trap here is that Cloud Spanner's architecture requires splitting the hot row's key to distribute writes across splits, not just adding more nodes.

How to eliminate wrong answers

Option A is wrong because increasing the number of nodes in Cloud Spanner increases overall throughput and storage capacity, but does not resolve contention on a single hot row—that row is still managed by one split and one leader, so write latency remains high. Option B is wrong because interleaved tables co-locate parent and child rows for efficient joins and strong consistency, but they do not help with write contention on a single frequently updated row; they actually increase the likelihood of contention if the parent row is the hot row. Option D is wrong because migrating to Cloud Bigtable is not a recommended design pattern for this scenario; Bigtable handles hotspots via automatic sharding, but the question asks for a design pattern within Cloud Spanner, and Bigtable does not support global, strongly consistent transactions or SQL queries, which the application likely requires.

Full explanation →
530
MCQmedium

Refer to the exhibit. The Cloud Run service is experiencing high tail latency under moderate load. Which change would most effectively reduce latency?

A.Increase CPU limit to 2.
B.Increase containerConcurrency to 250.
C.Increase timeoutSeconds to 600.
D.Reduce containerConcurrency to 10.
AnswerD

Lower concurrency reduces request queuing per container, improving tail latency under load.

Why this answer

High tail latency under moderate load often indicates that requests are queuing behind each other due to excessive concurrency. Reducing `containerConcurrency` to 10 limits the number of simultaneous requests each container instance handles, which reduces queueing delay and improves per-request response time. This is the most effective change because it directly controls the request multiplexing level, preventing a single instance from being overwhelmed.

Exam trap

The PCD exam often tests the misconception that increasing resources (CPU/memory) or timeouts always improves performance, when in fact controlling concurrency is the key to reducing tail latency in serverless platforms like Cloud Run.

How to eliminate wrong answers

Option A is wrong because increasing the CPU limit to 2 does not address the root cause of tail latency; it may reduce compute-bound delays but does not control request queuing or concurrency pressure. Option B is wrong because increasing `containerConcurrency` to 250 would exacerbate the problem by allowing more simultaneous requests per instance, increasing queueing and tail latency. Option C is wrong because increasing `timeoutSeconds` to 600 only extends the maximum request duration, which does not reduce latency; it may even mask underlying performance issues by allowing slow requests to linger longer.

Full explanation →
531
MCQhard

A company is migrating a legacy Java application to Cloud Run. The application requires a specific Java version (Java 11) and writes temporary files to disk. The application also uses a proprietary library that is not available in public repositories. The team has created a Dockerfile that installs Java 11, copies the JAR file, and sets the entrypoint. They are using Cloud Build to build the container and deploying to Cloud Run. The deployment succeeds, but when they send requests, the application fails with a "Permission denied" error when trying to write to /tmp. The team has verified that the Cloud Run service has the correct permissions via a service account. They have also checked that the filesystem is writable at /tmp by default. What is the most likely cause of the error?

A.Add a RUN chmod 777 /tmp command in the Dockerfile before the entrypoint.
B.Increase the memory limit of the Cloud Run service.
C.Change the base image to one that includes Java 11 and ensures the /tmp directory is writable.
D.Use a Cloud Storage FUSE mount for temporary storage.
AnswerC

A proper base image with the correct filesystem permissions resolves the issue.

Why this answer

The most common cause of write permission errors in Cloud Run is using a base image that does not have a writable /tmp directory or has the wrong filesystem permissions. Official base images like gcr.io/distroless/java or OpenJDK images are configured with a writable /tmp. Option A is incorrect because running chmod inside the Dockerfile may not fix the underlying issue if the filesystem is read-only or if the user doesn't have sufficient privileges; plus, Cloud Run may enforce security policies that prevent such operations.

Option B is incorrect because memory limits do not affect filesystem write permissions; they relate to memory allocation. Option D is incorrect because Cloud Storage FUSE is an overcomplication; the issue is simply the base image's /tmp configuration.

Full explanation →
532
MCQhard

A company uses Cloud SQL for PostgreSQL with a cross-region replica for disaster recovery. They need to perform a regional failover with minimal data loss. What configuration is required?

A.Use Database Migration Service instead
B.Promote the cross-region replica to a primary instance
C.Take a manual backup of the primary and restore to the replica
D.Enable synchronous replication
AnswerB

Promotion makes the replica a standalone instance, and data up to the promotion point is preserved.

Why this answer

Cloud SQL cross-region replicas use asynchronous replication, so some data loss is possible. To minimize loss, ensure that the replication lag is low and that automated backups are taken frequently. However, the key is to promote the replica to a standalone instance, which will stop replication and make it writable.

Full explanation →
533
MCQhard

A large e-commerce platform uses Cloud Bigtable to store user session data and product recommendations. They have a single cluster in a single zone. During a recent zone outage, the application became unavailable for 30 minutes because Cloud Bigtable was unreachable. The team needs to ensure high availability for the session data with a Recovery Time Objective (RTO) of less than 5 minutes and a Recovery Point Objective (RPO) of zero (no data loss). What should they do?

A.Migrate the session data to Cloud Memorystore for Redis with persistence and replication.
B.Add a second cluster in a different zone within the same region and use multi-cluster routing to automatically failover.
C.Configure replication to a second cluster in a different region and use global routing to failover.
D.Use Cloud Bigtable's single-cluster replication to a different zone.
AnswerB

Multi-cluster within region provides zone-level HA with fast replication.

Why this answer

Adding a second Cloud Bigtable cluster in a different zone within the same region and enabling multi-cluster routing provides automatic failover with an RTO of under 5 minutes and an RPO of zero. Multi-cluster routing directs read and write requests to the nearest healthy cluster, and replication between clusters is synchronous within a region, ensuring no data loss during a zone outage.

Exam trap

The PCD exam often tests the misconception that cross-region replication can achieve an RPO of zero, but candidates must remember that only intra-region replication is synchronous, while cross-region replication is asynchronous and introduces data loss risk.

How to eliminate wrong answers

Option A is wrong because Cloud Memorystore for Redis with persistence and replication does not guarantee an RPO of zero; asynchronous replication can lose recent writes during a failover, and it is not designed for the same throughput and latency characteristics as Cloud Bigtable for session data. Option C is wrong because configuring replication to a second cluster in a different region uses asynchronous replication, which cannot achieve an RPO of zero due to cross-region replication lag, and global routing introduces higher latency and potential data inconsistency. Option D is wrong because Cloud Bigtable does not support single-cluster replication; replication is always between two or more clusters, and the term 'single-cluster replication' is a misnomer that does not exist in Cloud Bigtable's architecture.

Full explanation →
534
MCQeasy

A company wants to monitor the CPU utilization of their Compute Engine instances and automatically trigger scaling actions if utilization exceeds 80% for 5 minutes. Which service should they use?

A.Managed instance group autoscaler
B.Cloud Monitoring
C.Cloud Scheduler
D.Cloud Load Balancing
AnswerA

Autoscaler uses monitoring metrics to trigger scaling actions.

Why this answer

Managed instance group (MIG) autoscaler is the correct service because it is designed to automatically adjust the number of Compute Engine instances based on configured utilization metrics. By setting a target CPU utilization of 80% over a 5-minute window, the autoscaler will add or remove instances to maintain that threshold, directly meeting the requirement for automatic scaling actions.

Exam trap

The PCD exam often tests the distinction between monitoring services (Cloud Monitoring) and action-oriented services (autoscaler), leading candidates to pick Cloud Monitoring because they confuse alerting with automatic scaling.

How to eliminate wrong answers

Option B is wrong because Cloud Monitoring is a monitoring and alerting service that collects metrics, logs, and events, but it does not perform automatic scaling actions; it can trigger alerts but not directly add or remove instances. Option C is wrong because Cloud Scheduler is a cron job service for scheduling tasks at specified times, not for reacting to real-time CPU utilization thresholds. Option D is wrong because Cloud Load Balancing distributes traffic across instances but does not monitor CPU utilization or trigger scaling actions; it works in conjunction with autoscalers but does not perform scaling itself.

Full explanation →
535
MCQmedium

A developer deploying a new version of a microservice sees a sudden increase in error logs in Cloud Logging. The errors are 500 responses from the service. What is the most efficient way to investigate the root cause?

A.Use Cloud Trace to view the trace of failed requests
B.Revert to the previous version immediately
C.Check the CPU and memory metrics in Cloud Monitoring
D.Analyze the error logs using Log Analytics and create a log-based metric
AnswerA

Cloud Trace records traces for each request, including errors, allowing you to see the exact step that failed.

Why this answer

Cloud Trace provides end-to-end latency data and can capture detailed spans for individual requests, including those that resulted in 500 errors. By filtering traces to failed requests, you can pinpoint the exact service or function call that caused the error, making it the most efficient root-cause investigation method without requiring code changes or additional instrumentation.

Exam trap

The PCD exam often tests the misconception that log analysis alone is sufficient for debugging distributed systems, but the trap here is that Cloud Trace provides request-scoped context that logs lack, making it the most efficient first step for 500 errors in a microservice deployment.

How to eliminate wrong answers

Option B is wrong because reverting immediately is a reactive rollback that does not identify the root cause; it may resolve symptoms but wastes time if the issue is not version-related. Option C is wrong because CPU and memory metrics show resource utilization but cannot reveal application-level logic errors, such as a null pointer exception or a failed database query, that cause 500 responses. Option D is wrong because analyzing error logs and creating a log-based metric is useful for monitoring trends but is less efficient for pinpointing the specific failing request path; Cloud Trace directly correlates traces with error status codes for faster diagnosis.

Full explanation →
536
Multi-Selecteasy

Which TWO of the following are valid ways to export Cloud Logging logs to BigQuery?

Select 2 answers
A.Use the Logging API to write logs directly to BigQuery
B.Use a Dataflow pipeline to stream logs from Pub/Sub to BigQuery
C.Create a log sink with destination set to BigQuery dataset
D.Use the BigQuery Data Transfer Service for Cloud Logging
E.Use Cloud Monitoring to send logs to BigQuery
AnswersB, C

This is a valid alternative path for exporting logs to BigQuery.

Why this answer

You can use a Dataflow pipeline to read Cloud Logging logs from a Pub/Sub topic (where logs are routed via a log sink) and stream them into BigQuery for real-time analysis. This is a common pattern for custom log processing and transformation before loading into BigQuery. Option C is correct because Cloud Logging allows you to create a log sink directly with a destination of a BigQuery dataset, which automatically exports logs in near real-time without additional infrastructure.

Exam trap

The PCD exam often tests the distinction between direct sink destinations (BigQuery, Pub/Sub, Cloud Storage) and indirect methods like Dataflow or custom code, leading candidates to mistakenly think the Logging API or BigQuery Data Transfer Service can be used for export.

Full explanation →
537
MCQmedium

A development team is using Cloud Build to deploy containerized applications to GKE. They want to ensure that only containers that have passed security scans and unit tests are deployed to production. Which approach should they use?

A.Deploy to a staging cluster first, then manually promote to production using kubectl.
B.Use Cloud Build with a multi-step pipeline that includes test and security scan steps, and only promote to production after successful completion.
C.Use Cloud Deploy to automate delivery with approval gates.
D.Configure Cloud Build triggers to deploy directly to production on every push.
AnswerB

This ensures that only containers that pass all checks are deployed, maintaining quality and security.

Why this answer

Using a multi-step Cloud Build pipeline with test and security scan steps, and then promoting to production only after success, ensures only validated containers are deployed. Direct deployment to production on every push is risky. Manual promotion defeats automation.

Cloud Deploy adds unnecessary complexity for this simple requirement. Thus, option B is correct.

Full explanation →
538
MCQmedium

A company runs a global stock trading platform that requires strong consistency across regions. They need to read the latest account balance after each trade. Which Cloud Spanner read mode should they use?

A.Strong reads
B.Read-only transactions with external consistency
C.Stale reads with bounded staleness of 10 seconds
D.Bounded staleness reads with max staleness of 1 second
AnswerA

Strong reads always return the most recent committed data with global consistency, required for financial transactions.

Why this answer

Strong reads provide the latest committed data with global consistency. Stale reads (bounded staleness) may return older data. External consistency is a Spanner property but not a read mode.

Full explanation →
539
MCQmedium

A company is planning to migrate from Snowflake to BigQuery. They have existing complex SQL transformations using Snowflake-specific functions like LISTAGG and PIVOT. What is the best approach to handle these differences?

A.Use BigQuery Data Transfer Service to automatically convert Snowflake functions.
B.Keep Snowflake for complex queries and use BigQuery only for simple analytics.
C.Use Cloud Spanner as an intermediary to translate queries.
D.Rewrite the SQL queries to use BigQuery equivalent functions and syntax.
AnswerD

Manual or tool-assisted rewriting is necessary as there is no automatic conversion.

Why this answer

Snowflake and BigQuery have different SQL dialects. The best approach is to use a migration assessment tool to identify incompatible syntax, rewrite queries to BigQuery-compatible syntax (e.g., use STRING_AGG instead of LISTAGG), and test thoroughly. BigQuery does not support PIVOT natively; use conditional aggregation.

Full explanation →
540
Multi-Selectmedium

A company is deploying a globally distributed user authentication service using Cloud Spanner. They need strong consistency for all reads and writes. Which two features should they use? (Choose two.)

Select 2 answers
A.Enable read-your-writes consistency
B.Use stale reads to reduce latency
C.Use strong reads (read timestamp = now)
D.Use mutations API with commit timestamps
E.Use bounded staleness reads
AnswersC, D

Strong reads return the latest committed data.

Why this answer

Strong reads (read timestamp = now) and mutations API with commit timestamp are used for strong consistency. Stale reads and bounded staleness are for weaker consistency. Read-write transactions also provide strong consistency, but the question asks for features.

The two correct are: strong reads and using commit timestamps for writes.

Full explanation →
541
MCQmedium

A company needs to store petabytes of time-series IoT sensor data and query it with single-digit millisecond latency at millions of reads per second. The data has a simple key-value structure with timestamps. Which Google Cloud database is MOST appropriate?

A.Firestore
B.BigQuery
C.Cloud Bigtable
D.Cloud Spanner
AnswerC

Correct: wide-column NoSQL, designed for time-series/IoT, single-digit ms latency, millions of QPS.

Why this answer

Cloud Bigtable is designed for petabyte-scale, low-latency, high-throughput NoSQL storage for time-series, IoT, and financial data. It scales horizontally by adding nodes. BigQuery is for analytics (seconds-to-minutes latency), Cloud SQL for OLTP (limited throughput), Firestore for document data.

Full explanation →
542
MCQhard

A Cloud Spanner instance is experiencing high latency for global secondary index queries. The team notices that queries on the secondary index are performing full table scans instead of index scans. What is the most likely cause?

A.The index is a local index without the STORING clause
B.The Spanner instance is under-provisioned
C.The index is a global index without the STORING clause, causing index join
D.The index is defined on a composite key that is not selective
AnswerC

Global indexes require a join with the base table to retrieve columns not stored in the index, leading to scans.

Why this answer

In Cloud Spanner, secondary indexes can be local (interleaved with parent) or global (non-interleaved). If an index is defined as a global index, queries might still need to join with the base table unless the index includes all needed columns. Without the `STORING` clause, Spanner must fetch additional columns from the base table, causing the query to scan the base table.

Full explanation →
543
MCQhard

You are a site reliability engineer for a fintech company that runs a latency-sensitive trading application on Google Kubernetes Engine (GKE). The application is instrumented with OpenTelemetry and exports traces and metrics to Cloud Monitoring and Cloud Logging. Recently, the team observed a gradual increase in p99 latency from 50ms to 500ms over the past week, and error rates have spiked to 5% from a baseline of 0.1%. You review the Cloud Monitoring dashboards and notice that the 'container/cpu/utilization' metric shows normal usage, but the 'container/memory/bytes_used' metric shows a steady climb, reaching 90% of the memory limit on several pods. The application logs contain many 'OutOfMemoryError' exceptions and 'GC overhead limit exceeded' messages. You also see that the HPA (Horizontal Pod Autoscaler) has not triggered any scale-up events because the 'custom/googleapis.com|container/cpu/utilization' metric is below the target utilization threshold. The cluster autoscaler is enabled and has sufficient node pool capacity. What is the most likely root cause and the best immediate action to resolve the issue?

A.Enable the Vertical Pod Autoscaler (VPA) in update mode to automatically adjust memory requests.
B.Switch the HPA to use the default 'container/cpu/utilization' metric instead of the custom metric.
C.Increase the memory request and limit for the pods to allow more memory usage.
D.Add a custom metric for memory utilization to the HPA and configure the target to scale when memory exceeds 70%.
AnswerD

This allows the HPA to react to memory pressure, scaling out pods to distribute memory load and reduce OOM errors.

Why this answer

The gradual memory increase and OutOfMemoryError exceptions indicate that the application is memory-bound, not CPU-bound. Since the HPA is configured to scale only on CPU utilization, it never triggers scale-up despite memory pressure. Adding a custom memory utilization metric to the HPA (option D) directly addresses the root cause by scaling pods when memory exceeds 70%, preventing OOM errors and reducing latency.

Exam trap

The PCD exam often tests the misconception that CPU is the only metric for HPA scaling, or that increasing resource limits alone solves memory pressure, when in fact memory-bound applications require scaling based on memory utilization to avoid OOM and latency degradation.

How to eliminate wrong answers

Option A is wrong because the Vertical Pod Autoscaler (VPA) adjusts resource requests/limits but does not scale the number of pods; it also cannot be used with HPA on the same metric, and update mode may cause pod restarts. Option B is wrong because switching to the default CPU metric would not help; CPU utilization is already normal, so the HPA would still not scale. Option C is wrong because simply increasing memory requests/limits without scaling out does not resolve the underlying issue of insufficient total memory capacity; pods will still hit the new limit eventually, and it does not address the latency spike caused by GC overhead.

Full explanation →
544
MCQeasy

To monitor Cloud SQL query performance, which metric should an engineer examine to see the rate of queries being executed?

A.database/disk/bytes_used
B.database/queries
C.database/memory/utilization
D.database/cpu/utilization
AnswerB

This metric shows query throughput.

Why this answer

The metric 'database/queries' tracks the number of queries executed per second. CPU/memory/disk utilization metrics are resource-oriented, not query-oriented.

Full explanation →
545
Multi-Selectmedium

A company needs to run HTAP (hybrid transactional/analytical processing) workloads on PostgreSQL. They require 4x faster OLTP performance than standard PostgreSQL and the ability to run analytical queries directly on transactional data without ETL. Which TWO Google Cloud databases should they consider? (Choose two)

Select 2 answers
A.Cloud Spanner
B.Cloud SQL for PostgreSQL
C.AlloyDB with BigQuery Omni
D.AlloyDB
E.BigQuery
AnswersC, D

AlloyDB combined with BigQuery Omni allows running analytical queries directly on transactional data without ETL, leveraging BigQuery's analytics engine. This satisfies HTAP requirements. Correct.

Why this answer

AlloyDB is a PostgreSQL-compatible database designed for HTAP workloads. It offers 4x faster OLTP performance compared to standard PostgreSQL and includes a built-in columnar engine that allows running analytical queries directly on transactional data without ETL. Therefore, AlloyDB alone (option D) satisfies the requirements.

Additionally, AlloyDB can be combined with BigQuery Omni (option C) to leverage BigQuery's analytical capabilities on AlloyDB data without ETL, providing another valid solution for HTAP. Cloud SQL for PostgreSQL (option B) and Cloud Spanner (option A) lack the columnar analytics engine, and BigQuery alone (option E) is not transactional. Hence, options C and D are correct.

Full explanation →
546
Multi-Selecthard

A company is using Cloud Monitoring to set up an SLO for a latency-sensitive API. They have defined a custom SLI: the proportion of requests with latency under 200ms. Which three components must they define to create a complete SLO configuration? (Choose three.)

Select 3 answers
A.A target (e.g., 99.9%)
B.An SLI definition with a good/bad time series
C.A burn rate alert policy
D.A metric threshold alert
E.A window of compliance (e.g., 30 days)
AnswersA, B, E

Correct: the desired success rate.

Why this answer

A target (e.g., 99.9%) defines the desired proportion of good events over a compliance window, which is essential for an SLO. In Cloud Monitoring, the target is the threshold against which the SLI is measured to determine if the SLO is met.

Exam trap

The PCD exam often tests that candidates confuse optional alerting policies (burn rate alerts, metric threshold alerts) with the mandatory components of an SLO configuration, which are strictly the SLI, target, and compliance window.

Full explanation →
547
MCQeasy

A developer needs to store session state for a user in a cloud-native application. Which storage solution is most appropriate?

A.Cloud SQL
B.Memorystore
C.Cloud Storage
D.Bigtable
AnswerB

Memorystore provides fast, in-memory caching for session data.

Why this answer

Memorystore (Redis) is the most appropriate solution for storing session state in a cloud-native application because it provides an in-memory data store with sub-millisecond latency, which is critical for fast session reads and writes. Session state is ephemeral, key-value data that requires high throughput and low latency, and Memorystore supports features like TTL (time-to-live) for automatic session expiration and persistence options for durability. This aligns with the cloud-native principle of stateless application tiers offloading state to a managed caching layer.

Exam trap

The PCD exam often tests the misconception that any managed database (like Cloud SQL or Bigtable) can handle session state, but the trap is that session state requires in-memory speed and automatic expiration, which only a caching solution like Memorystore provides, not disk-based or analytical databases.

How to eliminate wrong answers

Option A is wrong because Cloud SQL is a relational database designed for structured, transactional data with ACID compliance, not for high-speed ephemeral session state; its disk-based storage and connection overhead introduce latency unsuitable for frequent session lookups. Option C is wrong because Cloud Storage is an object store for blobs and files, not a low-latency key-value store; it lacks the sub-millisecond read/write performance and TTL-based expiration needed for session management. Option D is wrong because Bigtable is a wide-column NoSQL database optimized for analytical workloads with high throughput on large datasets, not for small, transient session records; its design for batch and streaming analytics makes it overkill and inefficient for per-request session operations.

Full explanation →
548
Multi-Selecthard

An engineer is designing a disaster recovery plan for a production AlloyDB cluster. The primary cluster is in us-central1. They need a cross-region replica that can be promoted to primary in the event of a regional failure. The replica should also be able to handle read traffic in the secondary region. Which THREE steps should they take? (Choose 3.)

Select 3 answers
A.Promote the secondary cluster to primary during a disaster.
B.Configure an instance-level read pool in the secondary cluster.
C.Enable AlloyDB Omni on the secondary cluster.
D.Create a secondary cluster in us-west1 using cross-region replication from the primary cluster.
E.Add a read pool to the secondary cluster to handle read traffic.
AnswersA, D, E

Promotion is the standard DR procedure for cross-region replication.

Why this answer

AlloyDB cross-region replication is achieved by creating a secondary cluster in another region with an external read pool. The secondary cluster can be promoted for failover. Instance-level read pool does not exist; read pools are cluster-level.

AlloyDB Omni is for on-premises, not cross-region replication.

Full explanation →
549
MCQhard

A company is migrating from MySQL 5.7 to Cloud SQL for MySQL 8.0. After migration, a legacy application that uses the 'GROUP BY' clause with implicit sorting stops working correctly. What is the most likely cause?

A.MySQL 8.0 changed the default collation, affecting string comparison.
B.MySQL 8.0 no longer supports the 'GROUP BY' clause.
C.The 'ONLY_FULL_GROUP_BY' SQL mode is enabled by default.
D.MySQL 8.0 removed the implicit sorting of GROUP BY results.
AnswerD

MySQL 8.0 no longer sorts GROUP BY results implicitly; an ORDER BY clause must be added to guarantee sort order.

Why this answer

MySQL 8.0 removed the implicit sorting of GROUP BY that was present in MySQL 5.7. Applications relying on this behavior must explicitly use ORDER BY to guarantee sort order.

Full explanation →
550
Multi-Selecteasy

A company deploys a microservice on Cloud Run and wants to minimize cold starts during traffic spikes. Which two steps should they take? (Select exactly 2.)

Select 2 answers
A.Enable CPU always allocated
B.Use Cloud CDN
C.Set max_instances to a high value
D.Set concurrency to 1
E.Set min_instances to a value greater than 0
AnswersA, E

CPU always allocated ensures instances are active and ready to serve requests immediately.

Why this answer

Enabling CPU always allocated (option A) prevents the CPU from being throttled when the container is not handling requests, which reduces cold start latency because the runtime environment remains warm and ready to process incoming traffic immediately. This is particularly effective for minimizing cold starts during traffic spikes because the container's CPU is always active, eliminating the need to spin up resources from a cold state.

Exam trap

The PCD exam often tests the distinction between scaling limits (max_instances) and proactive instance provisioning (min_instances), so the trap here is that candidates mistakenly think setting a high max_instances prevents cold starts, when in fact it only caps the maximum scale and does nothing to keep instances warm.

Full explanation →
551
Multi-Selecthard

A company has a multi-module repository. They want to build only the modules that have changes. Which two features can they combine to achieve this? (Choose 2)

Select 2 answers
A.Cloud Build queue
B.Build scripts that detect changes
C.Cloud Build substitutions
D.Cloud Build triggers with filepath filters
E.Cloud Source Repositories mirror
AnswersB, D

A build step can run a script (e.g., `git diff`) to identify changed modules and conditionally execute subsequent steps.

Why this answer

Build scripts can be written to compare the current commit hash against the previous build's commit hash, or use git diff to detect which files have changed, and then conditionally execute build steps only for the affected modules. This approach gives fine-grained control over the build process and can be integrated into any CI/CD pipeline.

Exam trap

The PCD exam often tests the distinction between features that trigger builds (like triggers with filepath filters) and features that manage build execution or configuration (like substitutions or queues), leading candidates to mistakenly select options that sound related but do not actually detect changes.

Full explanation →
552
MCQmedium

A company needs to run analytics on their Cloud SQL database without affecting OLTP performance. They want to use a read replica. What is the BEST practice?

A.Run analytics directly on the primary instance
B.Export the data to BigQuery periodically
C.Create a read replica in the same or different region and route analytics queries to it
D.Use Cloud SQL's HA instance for analytics
AnswerC

Read replicas allow offloading read queries without affecting the primary.

Why this answer

Using a cross-region read replica can offload analytics queries without impacting the primary instance, but for read scalability within region, an in-region replica works too.

Full explanation →
553
MCQmedium

An e-commerce company is migrating its on-premises MySQL database to Cloud SQL. They require synchronous replication to a standby instance in the same region for automatic failover with zero RPO. Which Cloud SQL configuration should they use?

A.Use Cloud Spanner instead, as it provides built-in HA across regions
B.Deploy two standalone Cloud SQL instances and configure MySQL native replication between them
C.Create a Cloud SQL for MySQL HA instance in the chosen region
D.Create a Cloud SQL for MySQL instance with a read replica in a different region
AnswerC

HA instances use synchronous replication to a standby in a different zone within the region, ensuring automatic failover and 0 RPO.

Why this answer

Cloud SQL HA instances use synchronous replication to a standby in a different zone within the same region, providing automatic failover and 0 RPO.

Full explanation →
554
MCQhard

A team is using Cloud Build to build a Go application. They want to cache Go module dependencies across builds to speed up builds. Which configuration should they add to cloudbuild.yaml?

A.Use a custom builder that pre-installs dependencies
B.Set up a bucket for caching and use substitutions
C.Use a Kaniko cache with a remote repository
D.Use Cloud Build's built-in caching feature by specifying a volume
AnswerB

Correct. You can use a Cloud Storage bucket as a persistent cache. By setting up a bucket and using substitution variables in cloudbuild.yaml, you can script steps to download the cached Go module directory before the build and upload it after. This speeds up builds by reusing previously downloaded modules.

Why this answer

Cloud Build volumes share data across steps within a single build, but they do not persist across separate builds. To cache Go module dependencies across builds, you must use an external storage service such as Cloud Storage. A common pattern is to set up a Cloud Storage bucket for caching and use substitution variables in cloudbuild.yaml to reference the bucket.

Steps can be added to download the cache before building and upload after building, using tools like gsutil.

Exam trap

The PCD exam often tests the understanding that Cloud Build volumes are ephemeral within a build, while persistent cross-build caching requires external storage like Cloud Storage. Candidates may mistakenly select volume-based caching or confuse Kaniko's image layer caching with application dependency caching.

How to eliminate wrong answers

Option A is wrong because using a custom builder that pre-installs dependencies does not leverage Cloud Build's native caching mechanism; it only shifts the dependency installation to a custom image, which still requires rebuilding the image for each dependency change and does not persist the cache across separate builds. Option B is wrong because setting up a bucket for caching and using substitutions is not a built-in Cloud Build feature for caching dependencies; while Cloud Storage can be used for artifact storage, it requires manual scripting to upload/download the cache and does not integrate with Cloud Build's volume-based caching. Option C is wrong because Kaniko cache with a remote repository is designed for caching container image layers, not for caching Go module dependencies; Kaniko is a tool for building container images, and its cache stores intermediate layers, not application-level dependency caches like Go modules.

Full explanation →
555
MCQmedium

A company needs to migrate a self-managed PostgreSQL database (version 12) to Cloud SQL for PostgreSQL with minimal downtime. They plan to use Database Migration Service (DMS). What is the correct set of steps?

A.Create a DMS connection profile for the source; create a migration job with 'Full dump + CDC'; configure source for logical replication; promote destination when ready.
B.Create a DMS connection profile for the source; create a migration job with 'Full dump' only; promote the destination after dump completes.
C.Create a DMS connection profile for the source; set up Cloud SQL for PostgreSQL; use pg_dump and pg_restore manually.
D.Create a DMS connection profile for the source; create a migration job with 'CDC' only; no initial dump is needed.
AnswerA

DMS continuous migration uses full dump then CDC (logical replication) to keep source and destination in sync with minimal downtime.

Full explanation →
556
Multi-Selectmedium

An organization is migrating a critical application to Google Cloud and needs to ensure high availability and disaster recovery. The application runs on Compute Engine and uses a stateful database. Which three design choices should they make? (Choose three.)

Select 3 answers
A.Use managed instance groups distributed across multiple zones.
B.Use regional persistent disks for the database.
C.Use a global load balancer to route traffic to the closest healthy region.
D.Use preemptible VMs to reduce costs for the database layer.
E.Deploy all instances in a single zone and use snapshots for backup.
AnswersA, B, C

MIGs across zones provide auto-healing and high availability.

Why this answer

Managed instance groups (MIGs) distributed across multiple zones provide automatic failover and self-healing for the Compute Engine instances. If a zone fails, the MIG automatically recreates instances in healthy zones, ensuring high availability for the application layer. This aligns with Google Cloud's best practices for regional resilience.

Exam trap

The PCD exam often tests the misconception that cost-saving measures like preemptible VMs can be applied to stateful workloads, but the trap here is that preemptible VMs are not guaranteed to run and thus cannot support a stateful database requiring persistent uptime and data integrity.

Full explanation →
557
MCQhard

A company is deploying a microservices architecture on GKE. They need to expose a set of related microservices under a single external IP address with path-based routing. Which Kubernetes resource should they use?

A.Service of type NodePort
B.NetworkPolicy
C.Service of type LoadBalancer
D.Ingress resource
AnswerD

Ingress provides path-based routing to multiple Services under one IP.

Why this answer

An Ingress resource is the correct choice because it provides HTTP/HTTPS layer-7 routing to expose multiple services under a single external IP address, using path-based or host-based rules. This directly meets the requirement of exposing a set of related microservices with path-based routing on GKE, whereas a Service of type LoadBalancer would create a separate external IP per service.

Exam trap

The trap here is that candidates often confuse a Service of type LoadBalancer with the ability to do path-based routing, but LoadBalancer only provides layer-4 TCP/UDP load balancing with a single external IP per service, not layer-7 path-based routing.

How to eliminate wrong answers

Option A is wrong because a Service of type NodePort exposes each service on a high-port on every node's IP, requiring clients to know the node IP and port, and does not provide a single external IP or path-based routing. Option B is wrong because a NetworkPolicy is a firewall rule that controls ingress and egress traffic between pods, not a mechanism for exposing services externally or routing traffic. Option C is wrong because a Service of type LoadBalancer provisions a separate external load balancer (and thus a separate external IP) for each service, failing the requirement to expose multiple services under a single IP with path-based routing.

Full explanation →
558
Multi-Selectmedium

A financial services company is migrating from Oracle to Cloud SQL for PostgreSQL. They need to ensure data consistency during the migration and minimize downtime. Which two actions should they take? (Choose 2)

Select 2 answers
A.Manually convert the Oracle schema to PostgreSQL using DDL statements.
B.Use DMS preview job to test migration before executing the full job.
C.Use Database Migration Service with continuous migration (CDC) to replicate ongoing changes.
D.Perform a full export using Oracle Data Pump and import directly into Cloud SQL.
E.Create connection profiles for both source and target databases in DMS.
AnswersC, E

CDC ensures near-zero downtime by replicating changes after the initial dump.

Why this answer

DMS continuous migration with CDC ensures minimal downtime by replicating changes. Connection profiles are necessary to configure source and target. Testing with DMS preview jobs is not a standard feature; DMS provides test jobs.

Manual schema conversion is not recommended; use Ora2Pg.

Full explanation →
559
MCQhard

A retail company processes customer orders through a pipeline. New orders are written to a Cloud Storage bucket as JSON files. A Cloud Function (currently triggered directly by Cloud Storage finalize events) parses the order and sends it to a third-party fulfillment service via an HTTP POST. As order volume grows, the team observes that the Cloud Function often times out (60s default) because the fulfillment service is slow. The team wants to decouple the processing to improve reliability. The order must be attempted at least once, and if the fulfillment service fails, retries should be exponential with a maximum of 5 attempts. Which solution should the team implement?

A.Use Cloud Tasks to create a queue that targets the Cloud Function. Configure the queue with exponential backoff and max retries of 5. Set the Cloud Function trigger to be HTTP instead of Cloud Storage.
B.Keep the Cloud Storage trigger, but increase the Cloud Function timeout to 540 seconds and add retry logic in the function code.
C.Use Pub/Sub notifications from Cloud Storage to a Pub/Sub topic, with a subscription that pushes to the Cloud Function. Enable dead letter topics for failed deliveries.
D.Replace the Cloud Function with a Cloud Run job that polls Cloud Storage for new files and sends orders to the fulfillment service. Use Cloud Scheduler to run the job every 5 minutes.
AnswerA

Cloud Tasks provides configurable retries, decouples the processing, and ensures at-least-once delivery. The HTTP-triggered function processes tasks from the queue.

Why this answer

Cloud Tasks provides the exact retry semantics required: exponential backoff and a maximum of 5 retry attempts. By using an HTTP-triggered Cloud Function and a Cloud Tasks queue, the order processing is decoupled from the Cloud Storage event, allowing the function to return quickly and rely on Cloud Tasks to handle retries if the fulfillment service is slow or fails. This ensures at-least-once delivery with controlled retries.

Option B is incorrect because increasing the timeout does not provide retries; the function would still block for up to 540 seconds without retrying on failure. Option C is suboptimal because Pub/Sub's push delivery retry configuration is less flexible than Cloud Tasks for HTTP endpoints, and it lacks built-in exponential backoff with a configurable max attempts without additional setup. Option D is inefficient because polling every 5 minutes introduces latency and does not provide real-time processing.

Full explanation →
560
Multi-Selectmedium

An organization needs to secure Cloud SQL for MySQL instances. Which two measures should they implement to restrict network access? (Choose TWO.)

Select 2 answers
A.Enable SSL enforcement (require_ssl flag)
B.Use IAM database authentication
C.Assign a public IP and configure authorized networks
D.Disable automatic storage increase
E.Use Private IP (VPC-only connectivity)
AnswersA, E

SSL enforcement ensures encrypted connections and is a security measure.

Why this answer

Using Private IP restricts access to the VPC network, and enabling the 'require_ssl' flag ensures encrypted connections. IAM authentication controls user access but does not restrict network access. Public IP with authorized networks allows network access from authorized IP addresses but is still public.

VPC Service Controls can restrict access, but the option is not listed; Private IP is the correct VPC-centric method.

Full explanation →
561
MCQhard

A company is migrating a PostgreSQL application to AlloyDB for improved OLTP performance and the ability to run analytical queries on the same data. Which feature of AlloyDB BEST supports this hybrid HTAP workload?

A.In-database ML inference
B.Cross-region replication
C.Columnar engine
D.Automatic storage scaling
AnswerC

The columnar engine accelerates analytical queries on transactional data, enabling HTAP.

Why this answer

AlloyDB's columnar engine is specifically designed to accelerate analytical queries by storing data in a column-oriented format, which allows for faster scans and aggregations on large datasets. This enables the same database to handle both OLTP (row-oriented) and OLAP (column-oriented) workloads without requiring a separate analytics system, directly supporting the hybrid HTAP requirement described in the question.

Exam trap

Some might mistakenly think that automatic storage scaling or replication features can address performance for mixed workloads, when in fact the columnar engine is the dedicated HTAP enabler in AlloyDB.

How to eliminate wrong answers

Option A is wrong because in-database ML inference is a feature for running machine learning models within the database, not a mechanism for improving analytical query performance on transactional data. Option B is wrong because cross-region replication is a disaster recovery and high availability feature that replicates data to another region, not a performance optimization for HTAP workloads. Option D is wrong because automatic storage scaling handles capacity management by adding storage as needed, but it does not change the storage format or query execution engine to support analytical queries.

Full explanation →
562
Drag & Dropmedium

Drag and drop the steps to troubleshoot a failed Cloud Build in the correct order.

Drag steps to the numbered slots on the right, or tap a step then tap a slot.

Steps
Order
1Step 1
2Step 2
3Step 3
4Step 4

Why this order

Troubleshooting involves reviewing logs and configuration, then fixing and re-running.

Full explanation →
563
MCQmedium

You are deploying a Cassandra database on GKE. Which resource type should you use to ensure stable network identities and persistent storage per pod?

A.Job with persistent disk
B.Deployment with persistent volume claims
C.StatefulSet with volumeClaimTemplates
D.DaemonSet with hostPath volumes
AnswerC

Provides stable identities and persistent storage.

Why this answer

StatefulSet is the correct resource because it provides stable, unique network identities (via headless Service and ordinal pod names) and persistent storage per pod through volumeClaimTemplates, which dynamically create PersistentVolumeClaims for each replica. This is essential for Cassandra, which requires stable node identities and dedicated storage to maintain cluster state and data consistency.

Exam trap

The PCD exam often tests the misconception that Deployments can handle stateful workloads by attaching PersistentVolumeClaims, but the trap is that Deployments lack stable network identities and per-pod storage binding, which are required for databases like Cassandra to maintain cluster membership and data integrity.

How to eliminate wrong answers

Option A is wrong because a Job is designed for batch processing tasks that run to completion, not for long-running stateful applications like Cassandra, and it does not provide stable network identities or persistent storage per pod. Option B is wrong because a Deployment provides replicas with ephemeral identities (random pod names) and shared PersistentVolumeClaims, which cannot guarantee stable network identities or dedicated storage per pod, leading to data conflicts and cluster instability for Cassandra. Option D is wrong because a DaemonSet runs one pod per node, which does not provide stable per-pod identities or dedicated persistent storage; hostPath volumes tie data to a specific node, causing data loss if the pod is rescheduled to a different node, and they lack the dynamic provisioning needed for Cassandra.

Full explanation →
564
Multi-Selectmedium

Which TWO statements are true about Cloud Functions? (Choose two.)

Select 2 answers
A.Cloud Functions automatically scale based on the number of incoming events
B.Cloud Functions require you to provide a container image from Container Registry
C.Cloud Functions are stateless by design
D.Cloud Functions can only be triggered by HTTP requests
E.Cloud Functions have a maximum execution timeout of 9 minutes for all runtimes
AnswersA, C

Cloud Functions scale out to handle multiple events concurrently.

Why this answer

Cloud Functions automatically scale horizontally based on the number of incoming events, such as HTTP requests or Pub/Sub messages. The platform handles this scaling transparently, spinning up new function instances as needed to handle concurrent invocations, and scaling down to zero when idle. This is a key serverless characteristic that eliminates the need for manual capacity planning.

Exam trap

The PCD exam often tests the misconception that Cloud Functions are limited to HTTP triggers or that the 9-minute timeout applies to all runtimes and trigger types, when in fact event-driven functions have a significantly longer timeout and multiple trigger options exist.

Full explanation →
565
MCQhard

A company is migrating a critical on-premises PostgreSQL database to Cloud SQL. They need to test the migration with production data shape without affecting the source. What is the recommended testing strategy?

A.Run shadow writes from the application to both source and Cloud SQL concurrently.
B.Export the production data using pg_dump and import into a test Cloud SQL instance.
C.Use Cloud SQL's clone feature to create a copy of the source after migration.
D.Use Database Migration Service to set up a continuous migration job and perform a dry run cutover.
AnswerD

DMS allows creating a migration job that can be stopped before cutover; this provides a replica with continuous sync for testing.

Why this answer

Database Migration Service (DMS) supports continuous replication from the source PostgreSQL database to a Cloud SQL instance. By setting up a migration job and performing a dry run cutover, you can validate the entire migration process—including schema conversion, data consistency, and application compatibility—using actual production data without affecting the source database. This approach is recommended because it tests the end-to-end workflow and minimizes risk.

Full explanation →
566
MCQmedium

A team is deploying a containerized application to Cloud Run. The application needs to process large files (up to 500 MB) uploaded by users. Which storage approach should they use to avoid Cloud Run's request size limit?

A.Upload files directly to Cloud Run
B.Mount a Cloud Filestore volume
C.Use Cloud Functions as a proxy for uploads
D.Use signed URLs to upload to Cloud Storage
AnswerD

Signed URLs enable direct client-to-Cloud Storage uploads, bypassing Cloud Run's request size limit.

Why this answer

Cloud Run has a default request size limit of 32 MB, which is far below the 500 MB files the application needs to process. Using signed URLs allows users to upload files directly to Cloud Storage, bypassing Cloud Run entirely. This approach avoids the request size limit and offloads the storage and retrieval of large files to a scalable, managed service.

Exam trap

The PCD exam often tests the misconception that Cloud Run can handle large request payloads by scaling, but the actual trap is that Cloud Run's request size is hard-limited to 32 MB, and candidates may overlook the need to offload uploads to a dedicated storage service like Cloud Storage.

How to eliminate wrong answers

Option A is wrong because Cloud Run enforces a 32 MB request size limit (configurable up to 32 MB), so uploading files directly to the container would fail for files up to 500 MB. Option B is wrong because Cloud Filestore is a network file system for GCE and GKE, not supported as a mounted volume in Cloud Run; Cloud Run only supports mounting Cloud Storage buckets via FUSE or NFS-like volumes. Option C is wrong because Cloud Functions also has a request size limit (10 MB for HTTP triggers), so using it as a proxy would still be constrained and adds unnecessary latency and complexity without solving the size limitation.

Full explanation →
567
MCQeasy

A company runs a web application on Google Kubernetes Engine (GKE) that uses Cloud SQL for its database. The application is deployed via Cloud Build and uses a custom container image stored in Artifact Registry. Recently, the development team noticed that new deployments are failing with a 'CrashLoopBackOff' error in the pod logs. The logs indicate that the application cannot connect to the Cloud SQL instance. The application uses the Cloud SQL Proxy sidecar container to connect. The Cloud SQL Proxy configuration has not changed recently. The GKE cluster is in region us-central1, and the Cloud SQL instance is in us-central1 as well. The team verified that the Cloud SQL instance is running and accepting connections from authorized networks. They also confirmed that the service account used by the pod has the 'Cloud SQL Client' role. The application has been running stable for months until a recent GKE cluster upgrade. The deployment YAML uses environment variables for database connection. The Cloud SQL Proxy is deployed as a sidecar container with command: '/cloud_sql_proxy -instances=<instance-connection-name>=tcp:5432'. The pod logs show the proxy attempting to connect but timing out. The network team confirms that there are no firewall rules blocking the connection. The Cloud SQL instance has an authorized network entry for the GKE cluster's nodes' IP range. What is the most likely cause of the connection failure?

A.The Cloud SQL Proxy container is using an outdated version that is incompatible with the current Cloud SQL API.
B.The application is using a wrong database password in the connection string.
C.The pod's namespace does not have the Cloud SQL Proxy configuration correctly set up.
D.The Cloud SQL instance is configured with a private IP and the GKE cluster is VPC-native but the Cloud SQL Proxy is configured to connect via public IP.
AnswerD

The proxy defaults to public IP; with a private IP instance and no public IP, connection times out.

Why this answer

The Cloud SQL Proxy sidecar is configured to connect via public IP (using the instance connection name without the `-private-ip` flag), but the Cloud SQL instance is configured with a private IP. Since the GKE cluster is VPC-native, the pod can reach the private IP within the VPC, but the proxy is attempting a public IP connection, which fails because the instance's authorized networks only include the GKE nodes' IP range (public IPs of nodes), not the pod's private IP range. The recent GKE cluster upgrade may have changed the node's public IP or networking configuration, exacerbating the mismatch.

Exam trap

The PCD exam often tests the distinction between public and private IP connectivity for Cloud SQL Proxy, where candidates assume the proxy automatically uses the correct IP based on the instance configuration, but it requires an explicit flag to use private IP.

How to eliminate wrong answers

Option A is wrong because the Cloud SQL Proxy version compatibility with the Cloud SQL API is not a common cause of timeout errors; the proxy uses the same SQL protocol and API version regardless of minor version changes. Option B is wrong because the error is a connection timeout, not an authentication failure; a wrong password would result in an 'access denied' error, not a timeout. Option C is wrong because the Cloud SQL Proxy configuration is defined in the pod spec, not at the namespace level; the proxy's command-line arguments are set in the container definition, and the namespace does not affect proxy connectivity.

Full explanation →
568
MCQmedium

A team wants to monitor custom application metrics from a Compute Engine instance. They use the Cloud Monitoring agent. Which metric type should they use to report a gauge measurement like current memory usage?

A.histogram
B.delta
C.cumulative
D.gauge
AnswerD

Gauge metric type reports instantaneous values.

Why this answer

A gauge metric type is specifically designed to report a value that can arbitrarily increase or decrease over time, such as current memory usage. The Cloud Monitoring agent supports gauge metrics for point-in-time measurements, and they are reported as a single data point without any aggregation window, making them ideal for snapshot-like observations.

Exam trap

Google Cloud often tests the distinction between metric types by presenting a scenario where a value can go up or down, and candidates mistakenly choose cumulative because they associate it with 'total usage' over time, forgetting that cumulative metrics must be monotonically increasing.

How to eliminate wrong answers

Option A is wrong because histogram metrics are used to capture the distribution of values over a time window (e.g., request latency percentiles), not a single instantaneous value like current memory usage. Option B is wrong because delta metrics represent the change in a value between two time points (e.g., requests per second), but current memory usage is not a rate or difference; it is an absolute snapshot. Option C is wrong because cumulative metrics monotonically increase over time (e.g., total bytes sent), and memory usage can decrease, which violates the monotonic property required for cumulative metrics.

Full explanation →
569
MCQmedium

A company is migrating an on-premises Oracle database to Cloud SQL for PostgreSQL. They require a deployment that supports up to 30,000 transactions per second (TPS) and provides automatic storage scaling. Which instance configuration should they choose?

A.Standard machine type with HDD and storage auto-increase enabled
B.High-memory machine type with SSD and storage auto-increase enabled
C.Custom machine type with 4 vCPUs and 15 GB memory, SSD, and storage auto-increase enabled
D.Lightweight machine type with SSD and storage auto-increase disabled
AnswerB

High-memory instances provide more memory per core, improving performance for PostgreSQL. SSD ensures low latency, and storage auto-increase prevents out-of-disk issues.

Why this answer

Cloud SQL supports storage auto-increase regardless of machine type. For high TPS, a high-memory machine type (e.g., highmem) provides more memory per CPU, which can improve performance for database workloads. SSD is required for consistent performance.

HDD is not recommended for production.

Full explanation →
570
MCQmedium

A Cloud Spanner query that uses a secondary index is taking longer than expected. The index includes a large payload column. How can the query performance be improved?

A.Use a composite key instead of an index
B.Change the index to a global index
C.Remove the index and use a full table scan
D.Add the payload column to the index using the STORING clause
AnswerD

STORING clause stores the column in the index, avoiding a join to the base table.

Why this answer

Using the STORING clause in a secondary index stores the specified columns in the index, allowing queries to be satisfied from the index alone without a join back to the base table. This reduces read latency.

Full explanation →
571
MCQmedium

A company is migrating an on-premises MySQL database to Cloud SQL. They want to minimize downtime and ensure data consistency using a one-time migration. Which approach should they use?

A.Set up a Cloud SQL replica and promote it
B.Use BigQuery Data Transfer Service
C.Create a Cloud SQL instance and use Database Migration Service with continuous replication
D.Export the database using mysqldump and import into Cloud SQL
AnswerC

DMS replicates data with minimal downtime and ensures consistency.

Why this answer

Database Migration Service (DMS) supports one-time migration with continuous replication until cutover, minimizing downtime.

Full explanation →
572
MCQmedium

An organization is migrating a large Oracle database to Cloud Spanner using Database Migration Service. They need to convert stored procedures and functions from PL/SQL to Spanner's SQL dialect. What is the best approach?

A.Use Ora2Pg to convert PL/SQL to PL/pgSQL and deploy to Cloud Spanner.
B.Migrate the PL/SQL code as-is; Cloud Spanner supports Oracle PL/SQL.
C.Manually rewrite PL/SQL as Spanner stored procedures.
D.Refactor the application to handle business logic outside the database using client libraries.
AnswerD

Since Spanner does not support stored procedures, the best practice is to move the logic to the application layer.

Why this answer

Spanner does not support stored procedures; the recommended approach is to migrate business logic to the application layer using client libraries, which provides better scalability and maintainability.

Full explanation →
573
MCQhard

A company has a monorepo with multiple services. They want to only build and test the services that have changed in a given commit. Which Cloud Build feature should they use?

A.Build triggers with filepath filters
B.Use a custom builder image
C.Build triggers with branch filters
D.Use Cloud Build substitutions
AnswerA

Filepath filters (included_files/ignored_files) limit the trigger to specific paths, enabling per-service builds.

Why this answer

Cloud Build triggers with filepath filters allow you to specify glob patterns that match changed files in a commit. When a commit includes changes only to files matching the filter, the trigger runs; otherwise, it is skipped. This enables selective building and testing of only the services that have changed in a monorepo, avoiding unnecessary builds for unaffected services.

Exam trap

Google often tests the distinction between filepath filters (which conditionally trigger builds based on changed files) and branch filters (which trigger builds based on branch names), leading candidates to mistakenly choose branch filters when the requirement is to build only changed services.

How to eliminate wrong answers

Option B is wrong because a custom builder image customizes the build environment (e.g., pre-installed tools) but does not provide any mechanism to conditionally trigger builds based on which files changed. Option C is wrong because branch filters trigger builds on specific branches (e.g., main or feature branches) regardless of which files changed, so they cannot restrict builds to only changed services. Option D is wrong because Cloud Build substitutions are user-defined variables that can be passed into build steps, but they do not control trigger conditions or filter builds based on file changes.

Full explanation →
574
MCQeasy

A company is migrating an on-premises MySQL 5.7 database to Cloud SQL for MySQL 8.0. The database includes MyISAM tables. What is the required action for these tables?

A.Use a MySQL 5.7 compatible Cloud SQL instance
B.Keep MyISAM tables as is
C.Migrate to Cloud Spanner instead
D.Convert MyISAM tables to InnoDB
AnswerD

Cloud SQL requires InnoDB; MyISAM is not supported.

Why this answer

Cloud SQL for MySQL 8.0 does not support MyISAM as the default storage engine; MyISAM tables must be converted to InnoDB. InnoDB provides ACID compliance, row-level locking, and crash recovery.

Full explanation →
575
MCQmedium

A developer deploys a service on Cloud Run with a concurrency setting of 1. The service makes external API calls. Under heavy load, the service starts returning 503 errors. What is the most likely cause?

A.The container image is too large.
B.The Cloud Run service is hitting the maximum number of requests per second limit.
C.The API endpoint rate limits the requests.
D.Instance concurrency is too low causing request queuing and timeout.
AnswerD

Low concurrency forces many instances, potentially hitting max instances and causing 503s.

Why this answer

With concurrency set to 1, each Cloud Run instance can handle only one request at a time. Under heavy load, new requests must wait for an available instance, and if the queue wait time exceeds the request timeout (default 5 minutes, max 60 minutes), the requests are dropped with a 503 HTTP status. This is the most direct cause of the 503 errors, as the service cannot scale fast enough or the queue depth exceeds limits.

Exam trap

A common misconception is that 503 errors are always caused by external API rate limits or downstream dependencies. However, with Google Cloud Run, a low concurrency setting can cause request queuing and timeouts, leading to 503 errors from the service itself.

How to eliminate wrong answers

Option A is wrong because container image size affects cold start latency but does not directly cause 503 errors under load; large images may increase startup time but not request queuing. Option B is wrong because Cloud Run does not have a hard 'maximum requests per second' limit; it scales instances based on concurrency and CPU utilization, and 503s from request rate limits are not a native Cloud Run behavior. Option C is wrong because the external API rate limiting would cause errors from that API (e.g., 429 or 503 from the API), not from the Cloud Run service itself; the question states the service returns 503, implying the issue is within the Cloud Run deployment.

Full explanation →
576
MCQmedium

You are deploying a Node.js application on Cloud Run. The container image is stored in Artifact Registry. After deploying with gcloud run deploy, the revision fails with 'Container failed to start. Failed to start and then listen on the port defined by the PORT environment variable.' The application listens on port 8080 by default. The Dockerfile uses EXPOSE 8080. The Cloud Run service is configured with container port 8080. You have verified that the container starts locally using docker run -p 8080:8080. What is the most likely cause of the startup failure?

A.The application is hardcoded to listen on port 8080 but the Cloud Run environment variable PORT may override it to a different value.
B.The application is trying to bind to a privileged port.
C.The Cloud Run service is configured with container port 443 by default.
D.The container does not have a proper HEALTHCHECK instruction.
AnswerA

Cloud Run sets the PORT variable; the app must read it.

Why this answer

Cloud Run sets the PORT environment variable to 8080 by default, but if the application is hardcoded to listen on 8080 instead of reading PORT, it may fail if the variable is not set or incorrect. Option B is wrong because the DEFAULT port variable is not used. Option C is wrong because the container port is set correctly.

Option D is wrong because port 8080 is not privileged.

Full explanation →
577
MCQhard

A company is migrating from Oracle to Cloud SQL for PostgreSQL. They have a stored procedure that uses Oracle's DBMS_OUTPUT.PUT_LINE for logging. How should they convert this to PostgreSQL?

A.Use RAISE EXCEPTION to output messages.
B.Use dbms_output package (installed via extension) to convert.
C.Use DBMS_OUTPUT.PUT_LINE as is; PostgreSQL supports it via the orafce extension.
D.Use RAISE NOTICE to output messages.
AnswerD

RAISE NOTICE is the PostgreSQL equivalent of DBMS_OUTPUT.PUT_LINE for logging.

Why this answer

PostgreSQL does not have DBMS_OUTPUT. The equivalent is RAISE NOTICE, which sends messages to the client. Alternatively, they can log to a table using INSERT.

RAISE is the most direct replacement. RAISE EXCEPTION is for errors, and dbms_output is not available.

Full explanation →
578
MCQmedium

A company is migrating a MySQL database with MyISAM tables to Cloud SQL for MySQL. They want to ensure the target database uses InnoDB for better transactional support and crash recovery. What should they do?

A.Run ALTER TABLE table_name ENGINE=InnoDB on each MyISAM table before migration.
B.Set the default_storage_engine flag to InnoDB on the source MySQL instance.
C.Enable the skip_myisam flag in Cloud SQL to automatically convert MyISAM to InnoDB.
D.Use mysqldump with the --compatible=innodb option.
AnswerA

This converts the table storage engine to InnoDB on the source, ensuring the dump/import uses InnoDB.

Why this answer

MyISAM tables must be converted to InnoDB before or during migration. The ALTER TABLE statement is the standard way to change the storage engine in MySQL.

Full explanation →
579
MCQhard

A company uses Cloud SQL for PostgreSQL as its primary database. They want to query this data from BigQuery for analytics without moving the data. They also need to ensure that BigQuery queries see the most recent data (within seconds of changes). Which approach is most suitable?

A.Create a Cloud Function that queries Cloud SQL every 5 seconds and writes results to BigQuery.
B.Use BigQuery federated queries to create an external table connected to Cloud SQL.
C.Use a scheduled query in BigQuery that exports Cloud SQL data to Cloud Storage and loads into BigQuery.
D.Set up Datastream to continuously replicate data from Cloud SQL to BigQuery, then query BigQuery tables.
AnswerD

Datastream provides near real-time CDC replication to BigQuery, ensuring data freshness within seconds.

Why this answer

BigQuery federated queries via external tables can directly query Cloud SQL without data movement. However, for near real-time freshness, the best approach is to use Datastream to replicate changes from Cloud SQL to BigQuery in near real-time. Federated queries alone have higher latency (minutes) and are not suitable for sub-second freshness.

Full explanation →
580
MCQeasy

A company is designing a cloud-native application on Google Cloud that requires low-latency access to a global user base. The application serves static content and dynamic APIs. Which strategy best minimizes latency while maintaining high availability?

A.Deploy the application in a single region and use Cloud Interconnect for global access.
B.Use Cloud CDN to cache static content and deploy the API across multiple regions with global load balancing.
C.Use Cloud Armor to protect the application and rely on Google's backbone for low latency.
D.Store all content in Cloud Storage and serve directly from there.
AnswerB

Cloud CDN caches at edge locations, and multi-region deployment with global load balancing reduces latency for dynamic content.

Why this answer

It combines Cloud CDN for caching static content at edge locations worldwide, reducing latency for static assets, and deploys the dynamic API across multiple regions with global load balancing (using Google Cloud's global external HTTP(S) load balancer) to route users to the nearest healthy backend, minimizing latency for dynamic requests while ensuring high availability through regional redundancy.

Exam trap

The trap here is that candidates may think Cloud Interconnect or Cloud Armor alone can solve global latency, overlooking the need for edge caching and multi-region deployment to reduce physical distance and provide redundancy.

How to eliminate wrong answers

Option A is wrong because deploying in a single region forces all traffic to traverse potentially long distances, and Cloud Interconnect provides dedicated connectivity to Google's network but does not reduce geographic latency for global users; it also creates a single point of failure. Option C is wrong because Cloud Armor provides DDoS protection and WAF capabilities but does not reduce latency; relying solely on Google's backbone does not address the need for edge caching or multi-region distribution. Option D is wrong because serving all content directly from Cloud Storage without a CDN or global load balancing results in higher latency for users far from the storage region, and Cloud Storage alone does not provide dynamic API serving or global traffic management.

Full explanation →
581
MCQhard

A company runs a stateful application on Compute Engine with regional persistent disks. They want to achieve high availability with automatic failover in case of a zone failure. Which architecture meets these requirements?

A.Use Cloud Storage FUSE to mount a multi-regional bucket as a filesystem
B.Use standard persistent disks with scheduled snapshots to a multi-region bucket
C.Use zonal persistent disks with a managed instance group in a single zone
D.Use regional persistent disks with a managed instance group spanning two zones
AnswerD

Regional disks provide synchronous replication; instance group autohealing restarts VMs on failure.

Why this answer

Regional persistent disks synchronously replicate data across two zones, and when combined with a managed instance group (MIG) spanning those same two zones, the application can automatically fail over to the healthy zone if one zone fails. The MIG's autohealing and health-check mechanisms detect the failure and recreate instances in the surviving zone, while the regional PD remains accessible from either zone, ensuring high availability without manual intervention.

Exam trap

The trap here is that candidates confuse zonal persistent disks with regional persistent disks, or assume that snapshots (Option B) provide automatic failover, when in reality snapshots are for backup/DR and require manual recovery steps, not instant zone-failure recovery.

How to eliminate wrong answers

Option A is wrong because Cloud Storage FUSE mounts a multi-regional bucket as a filesystem, but Cloud Storage is an object store with eventual consistency (not POSIX-compliant), and it does not support synchronous replication or automatic failover for stateful applications requiring persistent disk semantics. Option B is wrong because standard persistent disks are zonal resources; scheduled snapshots to a multi-region bucket provide backup and disaster recovery, not automatic failover within minutes, and restoring from a snapshot requires manual or scripted steps, not instant failover. Option C is wrong because zonal persistent disks are tied to a single zone, and a managed instance group in a single zone cannot survive a zone failure—the instances and disks become unavailable simultaneously.

Full explanation →
582
MCQeasy

A developer wants to run integration tests against a Cloud Spanner emulator in a local development environment. What is the best way to set up the emulator?

A.Use a mock library instead of an emulator.
B.Use the Cloud Spanner API directly without an emulator.
C.Run the Cloud Spanner emulator Docker container and set the SPANNER_EMULATOR_HOST environment variable.
D.Install the emulator using `gcloud emulators spanner start`.
AnswerC

Docker ensures a consistent environment and easy setup.

Why this answer

The Cloud Spanner emulator is distributed as a Docker container, and the recommended way to run it locally is to start that container and set the `SPANNER_EMULATOR_HOST` environment variable (e.g., `localhost:9010`) so that the Spanner client library redirects all API calls to the emulator instead of the production service. This provides a fully functional, in-memory Spanner instance for integration testing without incurring costs or requiring network access to GCP.

Exam trap

The Google PCD exam often tests the distinction between `gcloud emulators` commands that exist (Datastore, Firestore, Pub/Sub, Bigtable) and those that do not (Spanner), leading candidates to incorrectly assume `gcloud emulators spanner start` is valid when the correct approach is to run the Docker container and set the environment variable.

How to eliminate wrong answers

Option A is wrong because using a mock library (e.g., mocking the Spanner client) would test only the mock's behavior, not the actual Spanner API interactions, and would miss integration issues like schema validation, transaction semantics, or query syntax errors that the emulator catches. Option B is wrong because calling the Cloud Spanner API directly without an emulator would require a live GCP project, incur costs, and introduce network latency and authentication overhead, which defeats the purpose of a local, isolated development environment. Option D is wrong because `gcloud emulators spanner start` is not a valid command; the `gcloud emulators` subcommand supports Datastore, Firestore, Pub/Sub, and Bigtable, but not Spanner — Spanner's emulator is only available as a Docker image.

Full explanation →
583
MCQmedium

An engineer is migrating a large on-premises SQL Server database to Cloud SQL for PostgreSQL. They need to continuously replicate changes during the migration to minimize downtime. Which Database Migration Service job type should they use?

A.Bulk migration
B.Scheduled migration
C.Continuous migration
D.One-time migration
AnswerC

Continuous migration includes CDC after the initial dump, allowing near-zero-downtime cutover.

Why this answer

DMS offers continuous migration jobs that perform a full dump followed by CDC, enabling near-zero-downtime cutover.

Full explanation →
584
MCQmedium

A company is migrating an on-premises PostgreSQL database to Cloud SQL. They need to ensure minimal downtime and continuous replication during the migration. Which approach should they use?

A.Use Database Migration Service with continuous replication.
B.Use a cross-region replica from an existing Cloud SQL instance.
C.Export the database using pg_dump and import into Cloud SQL.
D.Create a Cloud SQL read replica of the on-premises database.
AnswerA

DMS supports ongoing replication with minimal downtime.

Why this answer

Database Migration Service (DMS) supports continuous replication from source to Cloud SQL using native PostgreSQL replication (logical replication). It minimizes downtime. Import/export involves downtime.

Creating a read replica from on-prem is not directly supported. Cloud SQL does not support cross-region replicas from on-prem.

Full explanation →
585
Multi-Selecteasy

Which TWO statements about building container images for Google Cloud Run are correct? (Choose TWO.)

Select 2 answers
A.The container must use HTTP/1.1; HTTP/2 is not supported.
B.The container must respond to health checks on the same port as the main application.
C.The container must be stateless and not rely on local file system persistence.
D.The container must listen on port 8080 by default.
E.The container must run as root user.
AnswersB, C

Health checks are sent to the container's port; the app must respond with 200 OK.

Why this answer

Cloud Run requires the container to respond to health checks (e.g., HTTP GET /health) on the same port as the main application. Cloud Run uses the same port for both user requests and health check probes; if the health check fails, the container is restarted. This ensures the container is always ready to serve traffic.

Exam trap

The PCD exam often tests the misconception that Cloud Run mandates port 8080 or requires HTTP/1.1, but the actual requirement is that the container must respond to health checks on the same port as the main application and must be stateless.

Full explanation →
586
MCQhard

A company uses Cloud Build to build Docker images and push them to Artifact Registry. They want to ensure that only images built from a specific Cloud Build trigger are deployed to production. Which combination of steps should they implement?

A.Configure IAM roles to restrict push access to Artifact Registry
B.Configure Binary Authorization with a policy requiring attestation from Cloud Build
C.Use Cloud Deploy with a manual approval gate
D.Use Cloud Build's built-in approval mechanism
AnswerB

Binary Authorization can enforce that only images with a valid attestation from Cloud Build are deployed.

Why this answer

Binary Authorization enforces deployment-time policies that require images to have a valid attestation from an approved authority. By configuring Cloud Build to create an attestation for images built from the specific trigger, and setting a Binary Authorization policy that requires that attestation, only those attested images can be deployed to production. This directly ensures that only images from that trigger are used.

Exam trap

The trap here is that candidates often confuse IAM-based access control (Option A) with deployment-time policy enforcement, failing to realize that IAM cannot distinguish images built from different triggers once they are in the registry.

How to eliminate wrong answers

Option A is wrong because IAM roles control who can push to Artifact Registry, not which specific images (e.g., from a particular trigger) are allowed to be deployed; it cannot enforce a policy based on the build source. Option C is wrong because Cloud Deploy with a manual approval gate adds a human review step but does not cryptographically verify that the image originated from a specific Cloud Build trigger; it relies on manual judgment, not automated attestation. Option D is wrong because Cloud Build's built-in approval mechanism only controls whether a build proceeds, not whether the resulting image is allowed to be deployed; it does not provide deployment-time enforcement.

Full explanation →
587
MCQmedium

A company is using Database Migration Service to migrate an on-premises PostgreSQL database to Cloud SQL for PostgreSQL. The source database is behind a firewall that does not allow direct connections from Google Cloud. What connectivity option should they use?

A.Use Cloud SQL Auth Proxy on a machine in the source network.
B.Use a Cloud VPN tunnel to connect the on-premises network to Google Cloud.
C.Set up VPC peering between the on-premises network and Google Cloud.
D.Configure a public IP for Cloud SQL and whitelist the source IP.
AnswerA

Cloud SQL Auth Proxy creates an encrypted tunnel and works over the internet, bypassing direct connectivity requirements.

Why this answer

Cloud SQL Auth Proxy provides a secure tunnel and can be run on the source network to connect to Cloud SQL. VPC peering requires direct network connectivity, which may not be possible through a firewall.

Full explanation →
588
MCQmedium

A company uses Cloud Build for CI/CD. They need to deploy a containerized app to Cloud Run automatically on every push to the main branch. Which Cloud Build configuration step is necessary?

A.Add a step to build the container with Dockerfile.
B.Add a step to run 'gcloud run deploy' command.
C.Add a step to push the image to Artifact Registry only.
D.Add a step to run unit tests.
AnswerB

This step performs the deployment to Cloud Run.

Why this answer

Cloud Build requires an explicit step to run the `gcloud run deploy` command in order to trigger a deployment to Cloud Run. While Cloud Build can build and push images, it does not automatically deploy to Cloud Run unless a deploy step is included in the build configuration. This step uses the built image (from Artifact Registry) and deploys it as a new revision to the specified Cloud Run service.

Exam trap

The PCD exam often tests the misconception that pushing an image to a registry automatically triggers a deployment, but in Cloud Build, each deployment must be explicitly commanded via a deploy step like `gcloud run deploy`.

How to eliminate wrong answers

Option A is wrong because building the container with a Dockerfile is necessary for creating the image, but it is not the step that deploys the app to Cloud Run; deployment requires an explicit deploy command. Option C is wrong because pushing the image to Artifact Registry only stores the image; it does not trigger a deployment to Cloud Run, which requires a separate deploy step. Option D is wrong because running unit tests is a quality assurance step that is optional and does not directly cause a deployment to Cloud Run.

Full explanation →
589
MCQhard

A DevOps team is designing a CI/CD pipeline for a microservices application deployed on Google Kubernetes Engine (GKE). They want to automatically build and deploy each service when a new tag is pushed to its repository. They also need to run integration tests against a staging environment before promoting to production. Which service should they use to orchestrate the pipeline?

A.Deployment Manager with a template
B.Cloud Deploy with a delivery pipeline
C.Cloud Build with a build trigger on tag push
D.Cloud Run with continuous deployment
AnswerC

Cloud Build can be triggered by a tag push, build the container, and deploy to GKE. It also supports running tests.

Why this answer

Cloud Build with a build trigger on tag push is the correct choice because it natively supports event-driven pipelines triggered by Git tags, enabling automatic build and deployment of each microservice. Combined with Cloud Build's ability to run integration tests in a staging environment and then promote to production, it provides a complete CI/CD solution for GKE without additional orchestration services.

Exam trap

Google Cloud often tests the distinction between CI/CD orchestration (Cloud Build) and delivery/rollout management (Cloud Deploy), leading candidates to choose Cloud Deploy because it sounds like a pipeline orchestrator, but it requires an external CI trigger to start the process.

How to eliminate wrong answers

Option A is wrong because Deployment Manager is an infrastructure-as-code tool for provisioning GCP resources, not a CI/CD pipeline orchestrator; it cannot trigger builds or deployments based on Git tag pushes. Option B is wrong because Cloud Deploy is a continuous delivery service that manages rollout strategies (e.g., canary, blue/green) but relies on an external CI system like Cloud Build to trigger the pipeline; it does not natively handle build triggers on tag push. Option D is wrong because Cloud Run with continuous deployment is designed for serverless container deployments, not for orchestrating multi-service CI/CD pipelines on GKE, and it lacks native support for tag-based triggers and staging-to-production promotion workflows.

Full explanation →
590
MCQmedium

A company wants to use BigQuery to query data stored in AWS S3 without copying it to Google Cloud. Which Google Cloud feature should they use?

A.Cloud Datastream
B.Cloud Storage FUSE
C.BigQuery Omni
D.BigQuery transfers
AnswerC

BigQuery Omni allows querying data directly in AWS S3 and Azure Blob.

Why this answer

BigQuery Omni enables querying data in S3 and Azure Blob Storage using a multi-cloud analytics approach.

Full explanation →
591
MCQmedium

A developer runs the command shown in the exhibit. They need to ensure that the application running on instance-3 can be restored quickly if it fails. What should they do?

A.Add instance-3 to an unmanaged instance group.
B.Configure Cloud DNS with a health check pointing to instance-3.
C.Create a managed instance group using an instance template based on instance-3's boot disk.
D.Take a snapshot of instance-3's disk and use it to create a new instance manually.
AnswerC

Managed instance groups automatically recreate failed instances.

Why this answer

A managed instance group (MIG) with an instance template based on instance-3's boot disk provides automated health checking, auto-healing, and managed instance replacement. If instance-3 fails, the MIG automatically recreates it from the template, ensuring rapid restoration without manual intervention. This aligns with the goal of designing highly available and reliable applications on Google Cloud.

Exam trap

The PCD exam often tests the distinction between health-check-based traffic routing (Cloud DNS) and health-check-based instance recovery (MIG auto-healing), leading candidates to confuse DNS failover with automated instance restoration.

How to eliminate wrong answers

Option A is wrong because an unmanaged instance group does not provide auto-healing or automated instance replacement; it is a static collection of instances that must be managed manually, so it cannot restore instance-3 quickly on failure. Option B is wrong because Cloud DNS with a health check only routes traffic away from an unhealthy instance but does not recreate or restore the instance itself; it is a traffic management solution, not a recovery mechanism. Option D is wrong because taking a snapshot and manually creating a new instance is a manual, slow process that does not meet the requirement for rapid, automated restoration; it lacks the automation and health-based recovery of a managed instance group.

Full explanation →
592
Multi-Selectmedium

A company is migrating a 500 GB MySQL database to Cloud SQL for MySQL 8.0. They want to test the migration with minimal risk before the final cutover. Which TWO testing approaches should they use?

Select 2 answers
A.Set up shadow writes to dual-write to both source and target databases.
B.Perform load testing using synthetic data only.
C.Run comparison queries on a snapshot of data from both databases.
D.Use Database Migration Service to migrate only the schema first.
E.Create a rollback plan before starting migration.
AnswersA, C

Shadow writes allow real-time testing by writing to both databases and comparing results.

Why this answer

Shadow writes allow testing by writing to both old and new DBs. Comparison queries validate data integrity. Load testing and rollback plan are also important but not testing approaches per se.

Full explanation →
593
MCQeasy

A company is deploying a static website on Cloud Storage with a custom domain. They want to serve the website over HTTPS. They have created a bucket with the same name as the domain and uploaded the files. They have verified the domain ownership in Search Console and added the bucket as a CNAME record in their DNS. Users report that when they navigate to the domain, they get a 404 error. The company has verified that the bucket's main page suffix is set to index.html. The team is confident the files are uploaded correctly. They need to resolve the 404 error and serve the site over HTTPS. What should they do?

A.Create a Cloud Load Balancer with the bucket as backend and update DNS.
B.Add a DNS A record pointing to the load balancer IP instead of a CNAME.
C.Enable Cloud CDN on the bucket.
D.Set the bucket's default object ACL to public read.
AnswerA

A Cloud Load Balancer provides SSL termination and a static IP for custom domains.

Why this answer

To serve a static website hosted on Cloud Storage with a custom domain over HTTPS, you must configure an external HTTP(S) load balancer with the bucket as the backend. The load balancer handles SSL termination and provides a static IP address or hostname. You then update your DNS to point the custom domain to the load balancer's IP (using an A record) or to the load balancer's hostname.

Option B is incorrect because a DNS A record pointing directly to a load balancer IP still requires the load balancer; the CNAME is not the issue. Option C is incorrect because Cloud CDN alone does not enable HTTPS for a custom domain; it requires a load balancer or other origin. Option D is incorrect because making objects publicly readable does not provide HTTPS support; the bucket's static website feature only supports HTTP.

Full explanation →
594
MCQhard

You are configuring a Cloud SQL for PostgreSQL instance for high availability. The instance must have automatic failover in under 60 seconds and avoid any data loss. Which configuration meets these requirements?

A.Create a regional HA configuration with synchronous replication
B.Use Cloud Spanner instead
C.Create a zonal HA configuration with asynchronous replication
D.Add a cross-region read replica and promote it on failure
AnswerA

Regional HA uses synchronous replication and provides fast failover.

Why this answer

Regional HA configures a primary and a standby in different zones within the same region, using synchronous replication to ensure no data loss and failover in under 60 seconds.

Full explanation →
595
MCQmedium

You are configuring a Cloud Monitoring alerting policy for a Cloud Run service. The service has a maximum of 10 concurrent requests per instance. You want to be alerted when the average number of concurrent requests per instance exceeds 8 for at least 1 minute. Which metric and condition type should you use?

A.Metric: run.googleapis.com/request_count, Condition type: Metric Threshold, Threshold: >8
B.Metric: run.googleapis.com/request_count, Condition type: Metric Absence, Duration: 1 min
C.Metric: resource/container/cpu/utilization, Condition type: Metric Threshold, Threshold: >80%
D.Metric: run.googleapis.com/request_count, Condition type: Change Rate, Threshold: >0.5
AnswerA

This metric measures active requests; threshold condition works for sustained high concurrency.

Why this answer

The `run.googleapis.com/request_count` metric tracks the number of concurrent requests per instance, which directly matches the requirement. A Metric Threshold condition with a threshold of >8 triggers an alert when the average exceeds 8 for at least 1 minute, aligning with the specified criteria.

Exam trap

The PCD exam often tests the distinction between metric types and condition types, where candidates confuse Metric Threshold (for sustained high values) with Change Rate (for sudden spikes) or Metric Absence (for missing data), leading to incorrect selections.

How to eliminate wrong answers

Option B is wrong because Metric Absence triggers when data is missing, not when a value exceeds a threshold; it would alert if the metric stops reporting, not when concurrent requests are high. Option C is wrong because `resource/container/cpu/utilization` measures CPU usage, not concurrent requests, and the threshold of 80% is unrelated to the request count. Option D is wrong because Change Rate detects sudden increases or decreases in the metric value, not a sustained high level; it would alert on a spike of >0.5 requests per minute, not when the average exceeds 8.

Full explanation →
596
Multi-Selecteasy

A company is designing a globally distributed application using Cloud Spanner. The application requires strong consistency and the ability to handle high read/write throughput. The team is concerned about inter-continental latency. Which two design choices would optimize performance while maintaining strong consistency? (Choose two.)

Select 2 answers
A.Enable leader-optimized routing to direct reads to the nearest leader region.
B.Use read-only replicas in each continent to serve reads locally.
C.Place a multi-region Spanner instance in geographic locations close to users.
D.Implement client-side caching with a short TTL for frequently accessed data.
E.Increase the number of nodes in the Spanner instance to improve throughput.
AnswersA, C

Leader-optimized routing reduces read latency while maintaining strong consistency.

Why this answer

A is correct because leader-optimized routing directs read requests to the nearest region that contains the leader replica for the requested data, reducing inter-continental latency while still reading from the leader to ensure strong consistency. C is correct because placing a multi-region Spanner instance in geographic locations close to users minimizes network round-trip time, and Spanner's synchronous replication across regions maintains strong consistency even with high read/write throughput.

Exam trap

The PCD exam often tests the misconception that read-only replicas or caching can provide strong consistency, but in Spanner, only leader replicas guarantee strong consistency, and any form of caching or stale replica reads breaks that guarantee.

Full explanation →
597
Multi-Selectmedium

A company uses Cloud Bigtable for a time-series application. They need to create a backup that can be restored to a different cluster in a different region for disaster recovery. Which THREE statements about Bigtable backups are correct?

Select 3 answers
A.The target cluster for restore must exist before the restore operation.
B.Backups can be used to migrate data across projects.
C.Backups are performed at the cluster level and include all tables in the cluster.
D.A backup can be restored to a different cluster in a different region.
E.Backups are incremental, capturing only changes since the last backup.
AnswersA, C, D

You must create the target cluster before initiating a restore.

Why this answer

When restoring a Cloud Bigtable backup, the target cluster must already exist in the specified region. The restore operation does not create a new cluster; it populates an existing cluster with the backed-up data. This ensures the cluster's configuration (e.g., node count, storage type) is predefined and meets the recovery requirements.

Exam trap

A common pitfall here is the misconception that Bigtable backups are incremental or can be used across projects, when in fact they are full snapshots restricted to the same project.

Full explanation →
598
Multi-Selecthard

A company runs a global e-commerce site on Cloud Spanner. They notice that some queries on the Orders table by customer_id are slow because customer_id is not part of the primary key (order_id). They want to avoid full table scans. Which TWO actions will improve query performance? (Choose two)

Select 2 answers
A.Create a secondary index on customer_id
B.Increase the number of nodes to improve scan speed
C.Re-create the table with customer_id as the first part of the primary key
D.Use interleaved tables to store customer data with orders
E.Use the STORING clause to include frequently accessed columns in the index
AnswersA, E

A secondary index allows fast lookup by customer_id.

Why this answer

A secondary index on customer_id will speed up queries. The STORING clause can include additional columns to avoid index joins. A global index is needed because customer_id is not in the primary key.

Interleaving would require a specific parent-child relationship.

Full explanation →
599
MCQmedium

A company is migrating a legacy monolithic application to Google Cloud. They want to minimize code changes and operational overhead while improving scalability. The application currently uses a relational database and stores user-uploaded images on a local filesystem. Which combination of Google Cloud services should they use?

A.Cloud Spanner for the database and Cloud Storage for images
B.Cloud Firestore for the database and Cloud Storage with Cloud CDN for images
C.Cloud SQL for the database and Cloud Storage with Cloud CDN for images
D.Compute Engine with attached SSD persistent disks for both database and images
AnswerC

Cloud SQL provides a managed relational database with minimal changes, and Cloud Storage with CDN serves images efficiently at scale.

Why this answer

Cloud SQL provides a fully managed relational database that requires minimal code changes when migrating from an existing relational database, while Cloud Storage with Cloud CDN handles user-uploaded images with scalable object storage and low-latency content delivery. This combination minimizes operational overhead by eliminating the need to manage database servers or file servers, and improves scalability through automatic replication and global edge caching.

Exam trap

Google Cloud often tests the misconception that any fully managed database (like Spanner or Firestore) is suitable for a legacy relational migration, but the trap here is that candidates overlook the requirement to minimize code changes and choose a NoSQL or globally distributed database that forces significant application rewrites.

How to eliminate wrong answers

Option A is wrong because Cloud Spanner is a globally distributed, horizontally scalable database designed for high-availability and strong consistency across regions, which introduces unnecessary complexity and cost for a legacy application that likely does not require global scale; it also requires significant code changes to adapt to Spanner's specific SQL dialect and consistency model. Option B is wrong because Cloud Firestore is a NoSQL document database that does not support relational queries, joins, or ACID transactions in the same way as a traditional relational database, forcing major application rewrites. Option D is wrong because Compute Engine with attached SSD persistent disks requires manual management of database software, backups, and failover, and does not provide the scalability or operational simplicity of managed services; it also lacks a CDN for image delivery, leading to higher latency and operational overhead.

Full explanation →
600
Multi-Selecteasy

A company needs to synchronize sales transactions from an on-premises Oracle database to BigQuery for near-real-time analytics. They want a serverless solution that requires minimal operational overhead. Which TWO services should they consider?

Select 2 answers
A.Pub/Sub
B.Dataflow
C.Datastream
D.Cloud Scheduler
E.Cloud Functions
AnswersA, C

Datastream can publish to Pub/Sub, which can then be streamed into BigQuery via subscriptions, though direct streaming is also possible.

Why this answer

Pub/Sub is correct because it provides a serverless, highly scalable ingestion service that can receive change data capture (CDC) events from an on-premises Oracle database via a connector like Debezium or Oracle GoldenGate, and stream them into BigQuery for near-real-time analytics without managing servers. Datastream is correct because it is a serverless CDC service specifically designed to replicate data from sources like Oracle directly to BigQuery, handling schema mapping and minimal operational overhead.

Exam trap

Google Cloud exams often test the distinction between services that are serverless ingestion sources (Pub/Sub, Datastream) versus processing or compute services (Dataflow, Cloud Functions), leading candidates to pick Dataflow because it is associated with streaming, despite it not being the minimal-overhead solution for simple CDC ingestion.

Full explanation →
Page 7

Page 8 of 13

Page 9

All pages

12345678910111213

Practice PCD by domain

Target a specific domain to shore up weak areas.

Designing highly scalable, available, and reliable cloud-native applicationsBuilding and testing applicationsDeploying applicationsIntegrating Google Cloud services
See all domains with question counts →