Google Professional Cloud DevOps Engineer PCDOE Questions 976–987 | Page 14/14

976

MCQmedium

After deploying a new version of a Cloud Run service, the team notices an increase in 5xx errors. They want to quickly revert to the previous version while minimizing user impact. What is the recommended approach?

A.Set the minimum number of instances of the new revision to 0.

B.Redeploy the previous version from the container registry.

C.Modify the ingress settings to restrict traffic to the new revision.

D.Use Cloud Run's traffic management to set 100% of traffic to the previous revision.

AnswerD

Traffic splitting achieves instant rollback.

Why this answer

Cloud Run supports traffic splitting between revisions, allowing you to instantly route 100% of traffic to the previous revision without redeploying. This minimizes user impact because the rollback is immediate and does not require rebuilding or re-pulling container images. Option D is correct because it leverages Cloud Run's built-in traffic management feature for zero-downtime rollbacks.

Exam trap

Google Cloud often tests the misconception that you must redeploy or delete a revision to roll back, when in fact Cloud Run's traffic management allows instant, traffic-level rollbacks without any deployment action.

How to eliminate wrong answers

Option A is wrong because setting the minimum number of instances of the new revision to 0 does not stop traffic from reaching it; it only affects scaling behavior, and the revision would still serve requests if it receives traffic. Option B is wrong because redeploying the previous version from the container registry is unnecessary and slower; Cloud Run already retains the previous revision, so you can simply shift traffic to it. Option C is wrong because modifying ingress settings to restrict traffic to the new revision would block all incoming traffic to that revision, but it does not automatically route traffic to the previous revision, leaving users with no service until you manually adjust routing.

Full explanation →

977

MCQmedium

A company is migrating Oracle to Cloud SQL for PostgreSQL using Ora2Pg. For a column defined as NUMBER(10,2) in Oracle, what is the corresponding data type in PostgreSQL?

A.INTEGER

B.NUMERIC(10,2)

C.DOUBLE PRECISION

D.TEXT

AnswerB

NUMERIC(10,2) matches the precision and scale.

Why this answer

Option B is correct because Oracle's NUMBER(10,2) specifies a fixed-point number with up to 10 digits total and 2 digits after the decimal point. PostgreSQL's NUMERIC(10,2) is the direct equivalent, offering identical precision and scale semantics. Ora2Pg automatically maps NUMBER(p,s) to NUMERIC(p,s) to preserve exact numeric storage.

Exam trap

Cisco often tests the misconception that NUMBER(p,s) maps to DOUBLE PRECISION or FLOAT, because candidates confuse fixed-point and floating-point semantics, especially when they see 'NUMBER' and assume it's a generic numeric type.

How to eliminate wrong answers

Option A is wrong because INTEGER stores only whole numbers without decimal places, losing the fractional precision required by NUMBER(10,2). Option C is wrong because DOUBLE PRECISION is a floating-point type that can introduce rounding errors for exact decimal values, unlike the fixed-point NUMERIC. Option D is wrong because TEXT is a variable-length string type, completely unsuitable for storing numeric data with precision and scale constraints.

Full explanation →

978

MCQmedium

An engineer wants to ensure that an alert is escalated if not acknowledged within 5 minutes. Which feature of Cloud Monitoring can achieve this?

A.Incident management tool like PagerDuty.

B.Notification channel with escalation configuration.

C.Using a webhook notification channel.

D.Alerting policy with a condition that checks for acknowledgment.

AnswerB

Escalation channels in notification settings allow sending to different recipients after a delay.

Why this answer

Notification channels in Google Cloud Monitoring can be configured with an escalation policy that defines a sequence of recipients or notification methods to be contacted if an alert is not acknowledged within a specified time. By setting the escalation duration to 5 minutes, the system automatically escalates the alert to the next tier or channel if no acknowledgment is received, directly meeting the engineer's requirement.

Exam trap

Google Cloud often tests the distinction between alerting policy conditions (what triggers an alert) and notification channel escalation (what happens after the alert fires), leading candidates to incorrectly select option D because they confuse the condition's 'duration' with the escalation timeout.

How to eliminate wrong answers

Option A is wrong because PagerDuty is an external incident management tool that integrates with Cloud Monitoring via notification channels, but it is not a built-in feature of Cloud Monitoring itself; the question asks for a feature of Cloud Monitoring, not an external tool. Option C is wrong because a webhook notification channel simply sends HTTP POST requests to a specified URL when an alert fires; it does not inherently support escalation logic or acknowledgment tracking. Option D is wrong because an alerting policy condition defines the metric or log criteria that trigger an alert, not the escalation behavior after the alert fires; acknowledgment and escalation are handled by the notification channel's configuration, not the policy condition.

Full explanation →

979

Multi-Selectmedium

A company is using Cloud SQL for PostgreSQL and wants to improve read scalability for a reporting dashboard that executes complex aggregate queries. The reports can tolerate up to 5 minutes of data staleness. Which two actions should the team take?

Select 2 answers

A.Enable PostgreSQL query cache to speed up aggregate queries

B.Create read replicas in Cloud SQL and configure the reporting application to connect to the replicas

C.Increase the number of vCPUs on the primary instance to handle more queries

D.Configure a connection pool to reduce connection overhead

E.Use Database Migration Service to continuously replicate data to BigQuery for reporting

AnswersB, E

Read replicas offload read traffic and can serve stale data with minimal lag.

Why this answer

Creating read replicas in Cloud SQL offloads read traffic from the primary instance, allowing complex aggregate queries to run on replicas without impacting write performance. Since the reports can tolerate up to 5 minutes of staleness, the asynchronous replication lag inherent in Cloud SQL read replicas is acceptable, making this a cost-effective and scalable solution.

Exam trap

Cisco often tests the distinction between scaling reads via replicas versus scaling writes or optimizing connections, and candidates mistakenly choose connection pooling or vertical scaling as a solution for read-heavy workloads.

Full explanation →

980

Multi-Selectmedium

Your organization is adopting DevOps practices and needs to bootstrap a Google Cloud organization with multiple projects. You want to enforce consistent resource naming conventions and apply common organization policies across all projects. Which two services should you use together to achieve this?

Select 2 answers

A.Cloud Shell and Cloud Source Repositories

B.Cloud Deployment Manager and Cloud Audit Logs

C.Service Accounts and IAM roles

D.Organization Policies and Resource Manager folders

AnswersC, D

Incorrect: Service accounts and IAM control access, not resource naming or organization policies.

Why this answer

Organization Policies allow you to centrally constrain actions across all projects in the hierarchy, while Resource Manager folders let you group projects and apply policies consistently. Together, they enable you to enforce naming conventions (e.g., via a custom constraint) and common policies (e.g., disabling external IPs) across multiple projects without manual per-project configuration.

Exam trap

Google Cloud often tests the distinction between identity/access management (IAM) and organization-level policy enforcement, leading candidates to mistakenly choose Service Accounts and IAM roles when the question specifically asks for consistent naming conventions and common policies across projects.

Full explanation →

981

MCQeasy

Your Cloud SQL for MySQL instance is running low on storage. You need to increase the storage size without downtime. What should you do?

A.You cannot increase storage online; you must create a new instance.

B.Create a clone of the instance with larger storage and switch traffic.

C.Enable automatic storage increase and it will happen automatically.

D.Edit the instance and increase the storage size; it will be applied online.

AnswerD

Cloud SQL allows online storage increase without downtime.

Why this answer

Option D is correct because Cloud SQL for MySQL supports online storage increases. When you edit the instance and increase the storage allocation, the operation is applied without requiring a restart or causing downtime. The underlying infrastructure uses live migration and dynamic volume expansion to make the additional capacity available immediately while the instance continues serving traffic.

Exam trap

The trap here is that candidates may assume any storage change requires downtime (Option A) or that cloning is the only safe method (Option B), but Cloud SQL explicitly supports online storage increases for MySQL instances.

How to eliminate wrong answers

Option A is wrong because Cloud SQL for MySQL does allow online storage increases; you do not need to create a new instance. Option B is wrong because creating a clone and switching traffic introduces unnecessary complexity and potential downtime during the cutover, and it is not the recommended method for a simple storage increase. Option C is wrong because automatic storage increase is a feature that triggers when storage usage reaches a threshold, but it does not allow you to manually specify a larger size; it only adds storage automatically based on usage, and you may need to increase storage on demand without waiting for automatic triggers.

Full explanation →

982

MCQhard

A global financial services company requires a multi-region Spanner instance with an RPO of zero and an RTO of less than 5 seconds for a critical transactional workload. The application writes to a single region and reads from multiple regions. Which Spanner configuration should they choose?

A.A single-region configuration with read replicas in multiple zones

B.A two-region multi-region configuration (e.g., nam6)

C.A dual-region configuration with two read-write regions and one read-only region

D.A three-region multi-region configuration (e.g., nam-eur-asia1)

AnswerB

Two-region multi-region configs use synchronous replication across two regions, providing RPO=0 and RTO<5s. Nam6 (us-central1 and us-east1) fits this scenario.

Why this answer

For zero RPO and sub-5-second RTO, a multi-region configuration with synchronous replication is needed. Spanner multi-region configurations like nam6 use synchronous replication across a primary and secondary region (with both read-write replicas). The leader region handles writes, and read-only replicas are for reads.

A two-region multi-region config (e.g., nam6) provides synchronous replication (RPO=0) and automatic failover in <5 seconds. Three-region configs (nam-eur-asia1) have higher write latency due to longer distances.

Full explanation →

983

Multi-Selectmedium

A data pipeline writes 10 TB of streaming data daily into Bigtable. The row key is based on the device ID and timestamp in reverse order. Recent data is queried most frequently. Which three design choices optimize performance and cost? (Choose THREE.)

Select 3 answers

A.Use HDD storage to reduce cost per GB.

B.Use a single cluster with SSD storage.

C.Use reverse timestamp to make recent data first.

D.Pre-split the table to avoid write hotspotting during initial load.

E.Enable compression on column families to reduce storage footprint.

AnswersB, D, E

SSD provides low latency; single cluster reduces cost.

Why this answer

Option B is correct because a single cluster with SSD storage provides low-latency access for frequently queried recent data, which is critical for streaming workloads. SSDs offer consistent single-digit millisecond latency, while HDDs would introduce higher latency unsuitable for real-time queries. This balances performance and cost for the described use case.

Exam trap

Cisco often tests the misconception that HDD is acceptable for cost savings in high-throughput streaming workloads, ignoring the latency requirements for frequent recent data queries, and that a design choice already implemented (like reverse timestamp) can be selected again as an optimization.

Full explanation →

984

MCQeasy

Which GCP tool provides recommendations for rightsizing Compute Engine VMs?

A.Cloud Console Cost Table.

B.Cloud Monitoring.

C.Cloud Billing Reports.

D.Recommender.

AnswerD

Recommender provides rightsizing and other cost optimization recommendations.

Why this answer

The Recommender tool in GCP uses machine learning to analyze historical usage patterns and provides actionable recommendations, including rightsizing suggestions for Compute Engine VMs. It identifies underutilized VMs and suggests optimal machine types to reduce costs without compromising performance, making it the correct choice for this task.

Exam trap

The trap here is that candidates often confuse Cloud Monitoring's alerting capabilities with proactive recommendations, assuming that monitoring data alone provides rightsizing suggestions, but Recommender is the dedicated tool for that purpose.

How to eliminate wrong answers

Option A is wrong because Cloud Console Cost Table is not a GCP tool; it is a generic term and does not exist as a specific service for rightsizing recommendations. Option B is wrong because Cloud Monitoring focuses on collecting metrics, logs, and setting alerts, but it does not generate rightsizing recommendations; it provides data that could be used manually but lacks the automated analysis. Option C is wrong because Cloud Billing Reports provide cost breakdowns and usage trends, but they do not offer specific VM rightsizing suggestions; they are for historical cost analysis, not proactive optimization.

Full explanation →

985

MCQmedium

A Memorystore for Redis instance is running out of memory and evicting keys. The application requires critical keys to be retained. The team wants to implement an eviction policy that removes less recently used keys first. Which eviction policy should they configure?

A.noeviction

B.allkeys-lru

C.volatile-ttl

D.volatile-lru

AnswerD

volatile-lru evicts only keys with an expire set, using LRU among them, preserving keys without TTL.

Why this answer

The volatile-lru eviction policy is correct because it removes the least recently used keys among those that have an expiration (TTL) set, which allows the application to retain critical keys that are not given a TTL. This policy ensures that only volatile keys are candidates for eviction, protecting non-expiring keys from being removed when memory is exhausted.

Exam trap

Cisco often tests the distinction between allkeys-lru and volatile-lru, trapping candidates who see 'least recently used' and immediately choose allkeys-lru without reading the requirement that critical keys must be retained.

How to eliminate wrong answers

Option A is wrong because noeviction prevents any key eviction and instead returns errors on write operations, which would cause the application to fail when memory is full rather than freeing space. Option B is wrong because allkeys-lru evicts the least recently used key from the entire keyspace, including critical keys that the application requires to be retained, which directly contradicts the requirement. Option C is wrong because volatile-ttl evicts keys with the shortest remaining TTL first, not the least recently used keys, so it does not match the requirement to remove less recently used keys.

Full explanation →

986

Multi-Selecthard

A company uses Cloud Spanner in a multi-region configuration. They want to perform a non-destructive disaster recovery drill to validate their ability to recover from a regional failure. Which THREE steps should they include in the drill? (Choose three.)

Select 3 answers

A.Validate that the restored instance has the correct data and schema.

B.Restore a Spanner backup to a new instance in a different region.

C.Update the disaster recovery runbook with any issues found during the drill.

D.Stop the primary Spanner instance to simulate a region failure and observe automatic failover.

E.Promote a cross-region read replica to a new primary instance.

AnswersA, B, C

Data integrity validation is critical to ensure the backup is usable and the recovery process works.

Why this answer

A non-destructive DR drill for Spanner should test the ability to restore from a backup to a different region (simulating recovery), validate the restored data integrity, and update runbooks based on lessons learned. Failing over a read replica is not applicable to Spanner (it uses multi-region replication, not read replicas). Stopping the primary instance is destructive.

Changing leader region is a configuration change that may affect live traffic.

Full explanation →

987

Multi-Selectmedium

Which TWO tools should be used for real-time incident collaboration and communication?

Select 2 answers

A.Google Meet only.

B.Jira.

C.Cloud Monitoring Incident Response (beta).

D.Google Chat with a dedicated incident room.

E.Cloud Trace.

AnswersC, D

This tool centralizes incident management and collaboration.

Why this answer

Cloud Monitoring Incident Response (beta) provides a dedicated interface for real-time incident management, including automated notifications, escalation policies, and a centralized timeline for collaboration. Google Chat with a dedicated incident room enables real-time communication and coordination among incident responders, allowing them to share updates, logs, and runbooks in a structured channel. Together, these tools fulfill the need for both incident orchestration and synchronous collaboration during active incidents.

Exam trap

Google Cloud often tests the distinction between tools that support real-time collaboration (like Chat rooms) versus tools that are for asynchronous tracking (like Jira) or monitoring-only (like Cloud Trace), leading candidates to select familiar but incorrect options like Jira for incident communication.

Full explanation →

Google Professional Cloud DevOps Engineer (PCDOE) — Questions 976–987