PCDE Practice Exam 52 — 20 Questions

Question 1

An SRE team uses Cloud Monitoring to alert on error budget burn rate. They configure a slow burn alert with a 6-hour lookback window and a burn rate factor of 5. What is the purpose of this slow burn alert?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

To detect a gradual increase in error rate that could exhaust budget over time. A slow burn alert with a long lookback window (6 hours) and a moderate burn rate factor (5) is designed to catch gradual, sustained increases in error rate that would exhaust the error budget over time if left unchecked. It complements fast burn alerts, which catch rapid spikes. The slow burn gives the team time to investigate and remediate before the budget is depleted.

Answer

To trigger when error budget is completely exhausted

Answer

To detect rapid spikes in error rate that could exhaust budget quickly

Answer

To calculate the remaining error budget

Question 2

An application running on Google Kubernetes Engine (GKE) uses OpenTelemetry SDK to export traces. Which Google Cloud service should receive these traces?

Accepted Answer

Cloud Trace. Cloud Trace is Google Cloud's distributed tracing service, purpose-built to ingest, store, and visualize trace spans exported by OpenTelemetry SDKs and other tracing libraries. When an application on GKE uses the OTel SDK to export traces, the correct backend is Cloud Trace, which natively supports the OTLP protocol and correlates spans across microservices. Cloud Trace is the only option designed specifically for trace data.

Answer

Cloud Logging

Answer

Cloud Monitoring

Answer

Error Reporting

Question 3

An SRE team has defined a service's availability SLI as the proportion of successful requests over a 5-minute window. They set an SLO of 99.9% over 30 days. What is the error budget for a 30-day period?

Accepted Answer

43 minutes 12 seconds. An SLO of 99.9% over 30 days means the allowed failure rate is 0.1%. Thirty days equals 43,200 minutes; 0.1% of that is 43.2 minutes, which converts to 43 minutes 12 seconds. This is the error budget — the total downtime or unsuccessful-request allowance the team can consume before breaching the SLO.

Answer

4 hours 19 minutes

Answer

7 minutes 12 seconds

Answer

43 minutes 12 seconds per week

Question 4

A Cloud Bigtable instance is experiencing high read latency. The team suspects hot spotting on a single node. Which tool should they use to identify the hotspot?

Accepted Answer

Key Visualizer. Key Visualizer is the correct tool because it is specifically designed to analyze access patterns in Cloud Bigtable and identify hot spotting—where a small range of row keys receives a disproportionate amount of traffic. It visualizes read and write heatmaps across row key space and time, allowing teams to pinpoint the exact keys causing the hotspot. Unlike general-purpose monitoring tools, Key Visualizer provides row-key-level granularity essential for diagnosing hot spotting.

Answer

Cloud Logging to review request logs

Answer

Bigtable Admin API to list tables

Answer

Cloud Monitoring (Stackdriver) to check CPU utilization

Question 5

A company is migrating its on-premises PostgreSQL database to Cloud SQL. The database is 2 TB and the migration must have minimal downtime. The source database supports continuous archiving. Which three steps should they take? (Choose THREE.)

Accepted Answer

Perform the cutover by promoting the Cloud SQL instance to primary.. Promoting the Cloud SQL instance to primary is the final step in a migration using continuous replication, which minimizes downtime by allowing the source database to remain operational until the cutover. This approach leverages Cloud SQL's ability to act as a replica that stays synchronized with the on-premises database via continuous archiving, ensuring data consistency with minimal interruption.

Answer

Use pg_dump to export the database and import into Cloud SQL.

Answer

Enable binary logging on the source database.

Question 6

A financial services company is deploying a new Cloud SQL for PostgreSQL instance that will handle sensitive customer data. The compliance team requires that all data at rest be encrypted with a customer-managed encryption key (CMEK) that the company can rotate and revoke independently. The company uses Cloud KMS and wants to ensure that the Cloud SQL instance uses the CMEK for all data, including backups and replicas. Which configuration should the database engineer implement?

Accepted Answer

Create a Cloud KMS key ring and key in the same region as the Cloud SQL instance, grant the Cloud SQL service account the Cloud KMS CryptoKey Encrypter/Decrypter role, and specify the key when creating the instance.. To use CMEK with Cloud SQL, the KMS key must be in the same region as the instance, and the Cloud SQL service account must have the CryptoKey Encrypter/Decrypter role. Specifying the key at instance creation ensures that all data, including backups and replicas, is encrypted with the CMEK. This meets the compliance requirement for customer-managed encryption.

Answer

Grant the Cloud SQL service account the Cloud KMS Admin role and specify the key during instance creation.

Answer

Enable CMEK by setting the `--disk-encryption-key` flag to the Cloud KMS key resource name when creating the instance using gcloud.

Answer

Create the Cloud SQL instance with default Google-managed encryption, then use Cloud KMS to encrypt the backups separately after creation.

Question 7

A company is migrating a MySQL database to Cloud SQL using Database Migration Service (DMS). The source database is on-premises with a public IP address. Which networking configuration is required on the source to allow DMS to connect?

Accepted Answer

Allowlist the DMS public IP addresses in the source database firewall.. When the source database is on-premises with a public IP, Database Migration Service (DMS) connects from Google Cloud over the public internet. To allow this, the source firewall must allowlist the DMS public IP addresses. This is the required networking configuration for DMS to reach the source.

Answer

Create a VPC peering connection between the source and DMS.

Answer

Install the Cloud SQL Auth Proxy on the source database.

Answer

Configure a Cloud VPN tunnel between the on-premises network and Google Cloud.

Question 8

A data engineer is writing a SQL query in BigQuery to calculate the running total of sales per product over time. The table 'sales' has columns product_id, sale_date, and amount. The result must include the cumulative sum ordered by sale_date for each product. Which SQL construct should be used?

Accepted Answer

SUM(amount) OVER (PARTITION BY product_id ORDER BY sale_date ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW). It uses a window function with a PARTITION BY clause to reset the running total per product and an ORDER BY with a ROWS frame to compute the cumulative sum over time. This is the standard SQL construct in BigQuery for calculating running totals within partitions.

Answer

GROUP BY product_id, sale_date with SUM(amount)

Answer

ROW_NUMBER() OVER (ORDER BY sale_date)

Answer

LAG(amount, 1, 0) OVER (ORDER BY sale_date)

Question 9

A company runs a GKE cluster and wants to ensure that during a planned node upgrade, their application remains available with minimal disruption. Which TWO steps should they take? (Choose two.)

Accepted Answer

Deploy multiple replicas of the application across different nodes.. Option A is correct because running multiple replicas spread across different nodes ensures that when a node is drained during the upgrade, other replicas on unaffected nodes continue serving traffic, maintaining application availability. Option D is correct because a PodDisruptionBudget with minAvailable guarantees that voluntary disruptions like node drains respect a minimum number of available pods, preventing GKE from evicting too many replicas at once. Option B is incorrect because max surge controls how many extra pods are created during a rolling update, not how pods survive node upgrades, and setting it to 100% does not protect availability during node drains. Option C is incorrect because setting the cluster autoscaler scale-down delay to 0 would make the cluster more aggressive about removing nodes, increasing disruption rather than minimizing it. Option E is incorrect because using a single replica per node reduces redundancy and would cause outages when that node is drained.

Answer

Increase the max surge of the deployment to 100%.

Answer

Set the cluster autoscaler scale-down delay to 0.

Answer

Use a single replica per node to simplify management.

Question 10

An e-commerce platform uses Cloud SQL for PostgreSQL. They need to run complex reporting queries that join several tables. These queries are slowing down the transactional workload. What should they do?

Accepted Answer

Use read replicas to offload reporting queries.. Read replicas in Cloud SQL for PostgreSQL allow you to offload read-only queries, such as complex reporting joins, from the primary instance. This separation reduces contention for CPU, memory, and I/O resources, preserving transactional performance. Read replicas are asynchronous and can handle heavy analytical workloads without impacting the primary's write path.

Answer

Create materialized views for common reports.

Answer

Change all joins to use subqueries.

Answer

Increase the number of vCPUs on the primary instance.

Question 11

A company uses Cloud Spanner in a multi-region configuration with the leader region in us-central1. They want to improve write latency for users in Europe. The application's writes are latency-sensitive and must be strongly consistent. Which action should the engineer take?

Accepted Answer

Change the leader region to a European region, such as europe-west1.. In Spanner multi-region, the leader region determines where writes are committed. To reduce write latency for European users, set the leader region to a European region (e.g., europe-west1). All writes will be committed in that region, reducing round-trip time for European clients. This does not affect consistency; Spanner still provides strong consistency globally. Adding more read replicas in Europe does not affect write latency. Moving the application to Cloud SQL or Bigtable does not provide the same global strong consistency or may not meet requirements.

Answer

Add more read-only replicas in the European region.

Answer

Use a regional Spanner instance in Europe and replicate data asynchronously to the multi-region instance.

Answer

Migrate the workload to Cloud Bigtable with replication.

Question 12

A company runs a critical application on Cloud SQL for PostgreSQL. The database engineer needs to ensure that if the primary instance fails, a standby instance in a different region can take over with minimal data loss. Which configuration should the Database Engineer implement?

Accepted Answer

Create a cross-region replica with asynchronous replication and manually promote it during a disaster.. Cloud SQL for PostgreSQL supports cross-region replicas with asynchronous replication, which allows a standby instance in a different region to be promoted manually during a disaster. This minimizes data loss by replicating changes asynchronously, though some transactions may be lost if the primary fails before replication completes. Automatic failover is not supported for cross-region replicas in Cloud SQL, so manual promotion is required.

Answer

Set up a second Cloud SQL instance and configure application-level dual-writes to both instances.

Answer

Configure high availability (HA) within the same region using a regional persistent disk.

Answer

Create a cross-region replica with synchronous replication and enable automatic failover.

Question 13

During an incident, the incident commander delegates tasks to multiple teams. Which communication model is MOST effective to reduce noise?

Accepted Answer

Use a single incident channel where all teams post updates.. A single incident channel consolidates all updates in one place, giving the incident commander and responders a shared timeline and reducing duplicate questions and fragmented context. This reduces noise because teams see each other's updates and avoid redundant status requests. It also creates a searchable record for post-incident review.

Answer

Use email for updates to avoid real-time noise.

Answer

Each team communicates in separate channels.

Answer

All updates go through the incident commander only.

Question 14

A team wants to improve incident response by creating a runbook for a common failure scenario: a database replication lag exceeds 5 seconds. Which Cloud Monitoring feature should be used to automatically trigger the runbook?

Accepted Answer

Configure a Cloud Monitoring alert to send a notification to a Cloud Function that executes the runbook steps. A Cloud Monitoring alerting policy can evaluate the replication lag metric and, when the threshold (5 seconds) is breached, send a notification to a Pub/Sub topic or webhook that triggers a Cloud Function to execute the runbook steps. This automates the response and ties directly to the metric condition.

Answer

Write a cron job on a VM that checks Cloud Monitoring metrics via API and runs the runbook

Answer

Create a Cloud Monitoring dashboard that displays replication lag and expect the on-call engineer to follow the runbook manually

Answer

Use Cloud Scheduler to run the runbook every 5 minutes

Question 15

A company is designing a monitoring strategy for a microservices application on Google Kubernetes Engine. They need to capture traces, metrics, and logs using a vendor-neutral approach. Which TWO components should they consider? (Choose 2)

Accepted Answer

OpenTelemetry Collector as a DaemonSet. Option A (OpenTelemetry Collector as a DaemonSet) is correct because the OpenTelemetry Collector is a vendor-neutral telemetry pipeline that can receive, process, and export traces, metrics, and logs, and deploying it as a DaemonSet ensures a collector pod runs on every GKE node to gather telemetry from all microservice pods. Option E (OpenTelemetry SDK for instrumentation) is correct because the OpenTelemetry SDK provides vendor-neutral APIs and instrumentation libraries that generate traces, metrics, and logs from the application code itself, which is the foundation for feeding data into the Collector. Option B (Cloud Logging agent) is not correct because it is a Google-specific logging agent, not a vendor-neutral solution for traces and metrics. Option C (Stackdriver agents on each node) is not correct because Stackdriver is the legacy Google-specific monitoring/logging agent, which is not vendor-neutral. Option D (Cloud Trace automatic instrumentation) is not correct because Cloud Trace is a Google Cloud-specific tracing service, so it does not meet the vendor-neutral requirement.

Answer

Cloud Logging agent

Answer

Stackdriver agents on each node

Answer

Cloud Trace automatic instrumentation

Question 16

A company uses Cloud Spanner with a multi-region configuration (nam3) for a global application. They notice that write latency has increased significantly during peak hours. After investigation, they find that the number of splits has increased from 10 to 50, and the CPU utilization on most nodes is below 10%. However, writes are being throttled due to excessive hot spots on a few nodes. What should they do?

Accepted Answer

Redesign the primary key to avoid monotonic increases. The symptom — many splits, low CPU, and write throttling from hot spots — is the classic signature of a monotonically increasing primary key (e.g., timestamp or sequential ID) that concentrates writes on the last split. Redesigning the primary key to distribute writes (e.g., using a hash prefix or UUID) eliminates the hot spot and restores write throughput.

Answer

Enable interleaved tables

Answer

Increase the number of nodes

Answer

Use a read replica

Question 17

Which TWO considerations are critical when migrating from Cassandra to Cloud Bigtable?

Accepted Answer

Denormalizing data to avoid joins. Option B is correct because Cloud Bigtable does not support joins, so data that was normalized across Cassandra tables must be denormalized into a single wide row or column family structure to preserve query access patterns. Option D is correct because Bigtable performance depends heavily on row key design; keys must be chosen to distribute reads and writes evenly across tablet servers and avoid hotspots, unlike Cassandra where partition keys and clustering columns handle distribution differently. Option A is incorrect because Cloud Bigtable does not support CQL; it uses the HBase API, Google Cloud CLI, or client libraries, so CQL queries must be rewritten. Option C is incorrect because Bigtable provides strong consistency for single-row operations, not the tunable eventual consistency model used in Cassandra. Option E is incorrect because Bigtable does not support secondary indexes; efficient filtering must be achieved through row key design or by scanning and filtering in the application.

Answer

Using CQL for queries

Answer

Maintaining eventual consistency model

Answer

Using secondary indexes for efficient filtering

Question 18

Your team uses PagerDuty for incident notifications. You need to configure Cloud Monitoring to send alerts to PagerDuty. Which notification channel type should you select?

Accepted Answer

PagerDuty. Cloud Monitoring natively supports PagerDuty as a notification channel type, allowing alerts to be routed directly to PagerDuty incidents. Selecting the PagerDuty channel and providing the integration key (service key) enables automatic incident creation and escalation. This is the correct, purpose-built integration for the scenario.

Answer

Slack

Answer

Cloud Pub/Sub

Answer

SMS

Question 19

Refer to the exhibit. The query used DATE_TRUNC(order_date, MONTH) as month. order_date is a TIMESTAMP column. What is the data type of the month column in the result?

Accepted Answer

TIMESTAMP. In BigQuery (the SQL engine for the PCDE exam), DATE_TRUNC with a TIMESTAMP input and MONTH granularity returns a TIMESTAMP value, not a DATE or DATETIME. The function truncates the timestamp to the first day of the month at 00:00:00 UTC, preserving the TIMESTAMP data type. Therefore, the month column in the result is of type TIMESTAMP.

Answer

STRING

Answer

DATE

Answer

DATETIME

Question 20

A company uses Cloud Bigtable for real-time analytics. They need to implement disaster recovery across regions with a recovery point objective (RPO) of no more than 5 minutes and recovery time objective (RTO) under 10 minutes. Which approach should they take?

Accepted Answer

Configure Bigtable replication with a secondary cluster in a different region and use Cloud DNS health checks to update routing policy on failover.. Cloud Bigtable replication lets you create a second cluster in another region with asynchronous replication, typically achieving sub-minute replication lag — well within the 5-minute RPO. Combined with Cloud DNS health checks and a failover routing policy, clients are redirected to the healthy cluster, meeting the sub-10-minute RTO. This is the native, purpose-built DR approach for Bigtable.

Answer

Perform daily exports of Bigtable data to Cloud Storage and import into a new cluster in case of failure.

Answer

Migrate the workload to Cloud Spanner multi-region for built-in disaster recovery.

Answer

Use Cloud Scheduler to trigger a script that takes a Bigtable snapshot every hour.