PCDE Practice Test 30 — 15 Questions

Question 1

Which of the following is an example of toil according to SRE principles?

Accepted Answer

Manually restarting failed pods in a Kubernetes cluster. Toil is defined in SRE as work that is manual, repetitive, automatable, tactical, devoid of enduring value, and scales linearly with service growth. Manually restarting failed pods in a Kubernetes cluster is a classic example of toil because it is repetitive, can be automated, and does not provide long-term value. The other options are not examples of toil.

Answer

Reviewing code from a junior developer

Answer

Writing a new feature for the application

Answer

Designing a new microservice architecture

Question 2

A team has an SLO of 99.9% availability over a 30-day month. They burn through their entire error budget in the first 10 days. Which of the following is the MOST appropriate immediate action according to SRE principles?

Accepted Answer

Freeze all feature releases and focus on reliability improvements. When error budget is exhausted, the team should halt all feature releases and focus on reliability improvements to prevent further degradation.

Answer

Reduce the error budget by lowering the SLO to 99%

Answer

Deploy a new feature to increase user engagement

Answer

Increase the SLO to 99.99% to act as a buffer

Question 3

A startup is building a mobile app with a relational database backend. They expect moderate traffic and need strong consistency, automatic backups, and point-in-time recovery. Which two Google Cloud database services meet these requirements? (Choose TWO.)

Accepted Answer

Cloud Spanner. Cloud Spanner (B) is correct because it is a fully managed relational database that provides strong consistency (external consistency) via TrueTime, and it supports automatic backups and point-in-time recovery (PITR) with configurable retention windows, making it suitable for a relational backend with moderate-to-high traffic. Cloud SQL (C) is correct because it is a managed relational database (MySQL, PostgreSQL, SQL Server) offering strong consistency, automated backups, and point-in-time recovery through binary logging/transaction logs, which fits a startup's moderate-traffic mobile app backend. Firestore (A) is a NoSQL document database, so it does not meet the relational requirement. Cloud Bigtable (D) is a NoSQL wide-column store optimized for high-throughput analytics and does not provide relational semantics or the required PITR/backup model for this use case. BigQuery (E) is a serverless analytics data warehouse for OLAP workloads, not a transactional relational backend.

Answer

Firestore

Answer

Cloud Bigtable

Answer

BigQuery

Question 4

An organization needs a multi-region database deployment with strong consistency and an RPO of zero in normal operation. They expect a regional outage and require automatic failover within seconds. Which database service and configuration meets these requirements?

Accepted Answer

Cloud Spanner multi-region configuration. Cloud Spanner multi-region configuration provides strong consistency (external consistency), an RPO of zero in normal operation, and automatic failover within seconds during a regional outage. It is designed for globally distributed, strongly consistent databases.

Answer

Cloud Firestore in multi-region mode

Answer

Cloud Bigtable with multi-cluster routing and replication

Answer

Cloud SQL with cross-region read replica

Question 5

A company uses Memorystore for Redis and wants to achieve high availability with automatic failover. They also need to periodically back up the data to Cloud Storage for disaster recovery. Which TWO features should they use? (Choose two.)

Accepted Answer

Standard Tier (replication). Standard Tier (replication) provides a primary-replica architecture with automatic failover, ensuring high availability. The Export functionality allows you to manually or scheduled export Redis data to Cloud Storage, enabling point-in-time recovery for disaster recovery purposes.

Answer

Redis Cluster

Answer

AOF persistence

Answer

Basic Tier

Question 6

A company is migrating an on-premises PostgreSQL database to Cloud SQL using Database Migration Service with continuous replication. The source database has binary logging enabled and uses the pglogical plugin. The migration job is failing after the full dump phase. Which THREE common issues should the engineer check? (Choose 3 correct answers.)

Accepted Answer

The source database is not configured for logical replication (e.g., wal_level is not logical).. Option A is correct because Database Migration Service for PostgreSQL relies on logical replication, which requires the source to have wal_level set to logical (plus max_replication_slots and max_wal_senders configured); if wal_level is not logical, the pglogical-based replication cannot stream changes after the initial dump. Option B is correct because continuous replication retains WAL segments until they are consumed by the replication slot, so insufficient disk space on the source for accumulating WAL files will cause the migration job to fail after the full dump phase. Option D is correct because the migration depends on a logical replication slot; if the slot was not created or has been exhausted/invalidated (e.g., due to exceeding max_slot_wal_keep_size), change data capture stops and the job fails. Option C is not applicable because Cloud SQL Auth Proxy is a client-side connector for reaching Cloud SQL instances, not something installed on the on-premises source for DMS replication. Option E is incorrect because the scenario explicitly states the source is PostgreSQL with pglogical, so it cannot be MySQL.

Answer

Cloud SQL Auth Proxy is not installed on the source.

Answer

The source database is MySQL.

Question 7

A retail company is designing a Cloud Spanner schema for an order management system. Orders are identified by a UUID and contain multiple line items. Each line item references a product. Which schema design best supports high read throughput for queries that retrieve all line items for a given order?

Accepted Answer

Create an Orders table and a LineItems table interleaved in Orders with ORDER_ID as the parent key.. Cloud Spanner interleaved tables store child rows (LineItems) physically adjacent to their parent row (Orders) on the same split, enabling a single key lookup to retrieve all line items for a given order without cross-table joins or distributed queries. This colocation maximizes read throughput by minimizing latency and avoiding scatter-gather operations across nodes.

Answer

Store orders and line items in a single table with repeated fields for line items.

Answer

Create separate Orders and LineItems tables with a foreign key relationship and index on ORDER_ID.

Answer

Denormalize product information into the LineItems table and store orders separately.

Question 8

A game company uses Cloud Bigtable to store player session data. Access patterns include looking up a player's most recent sessions and scanning sessions by time range. Which row key design is most appropriate?

Accepted Answer

Use a row key of player ID followed by reversed timestamp.. It leverages Cloud Bigtable's sorted row key structure to group all sessions for a given player ID together, while the reversed timestamp ensures that the most recent session appears first within that group. This design enables efficient point lookups for a player's latest sessions and supports range scans over time ranges by using the timestamp portion of the key.

Answer

Use only player ID as row key with column qualifiers for timestamps.

Answer

Prefix with timestamp and append player ID.

Answer

Use a hash of player ID as row key and store timestamps in cell versions.

Question 9

Which THREE of the following are best practices for designing BigQuery tables for business intelligence reporting?

Accepted Answer

Partition tables by a date or timestamp column used in WHERE clauses.. Option A is correct because partitioning a table by a DATE or TIMESTAMP column lets BigQuery prune partitions, so queries with WHERE filters on that column scan only the relevant partitions instead of the whole table, cutting cost and improving performance. Option D is correct because BigQuery is a columnar, denormalized analytics engine where nested repeated columns (ARRAY/STRUCT) let you store related child records inline, avoiding expensive joins and preserving hierarchical data for BI reporting. Option E is correct because clustering sorts data by the specified columns, so filters and GROUP BY operations on those columns benefit from block-level pruning and faster aggregation. Option B is wrong because splitting data into many small tables increases metadata overhead, complicates queries, and prevents BigQuery from efficiently pruning a single large dataset; partitioning/clustering is the recommended approach. Option C is wrong because normalization is a transactional-database practice; for BI reporting in BigQuery, denormalization (flattening with nested/repeated fields) is preferred to minimize joins and improve scan efficiency.

Answer

Store data in many small tables to reduce the amount of data scanned per query.

Answer

Normalize data to reduce data redundancy.

Question 10

You are setting up structured logging for an application running on Compute Engine. Your logs should include trace context and severity. Which TWO fields should you include in the JSON payload to enable correlation with Cloud Trace? (Select 2)

Accepted Answer

spanId. Option E (trace) is correct because Cloud Logging's structured logging special field `logging.googleapis.com/trace` (or the shorthand `trace`) carries the full resource name of the trace, e.g. `projects/PROJECT_ID/traces/TRACE_ID`, which is exactly what links a log entry to a trace in Cloud Trace. Option C (spanId) is correct because the `logging.googleapis.com/spanId` (shorthand `spanId`) field identifies the specific span within that trace, allowing the log entry to be correlated to the exact span in Cloud Trace. Together, trace and spanId provide the trace context needed for correlation. Option A (httpRequest) is not required here because it captures HTTP request metadata for logging, not trace correlation identifiers. Option B (message) is just the log payload text and does not provide trace context. Option D (severity) sets the log level (e.g., INFO, ERROR) and, while useful, is not a trace-correlation field.

Answer

httpRequest

Answer

message

Answer

severity

Question 11

A company wants to visualize request latency distribution across all services using a heatmap. Which Cloud Monitoring chart type should be used?

Accepted Answer

Heatmap

Answer

Scorecard

Answer

Line chart

Answer

Stacked bar chart

Question 12

A company uses Cloud SQL for PostgreSQL and needs to add a full-text search capability to a table with product descriptions. Which index type should be used?

Accepted Answer

GIN index. PostgreSQL supports full-text search using indexes created with the 'gin' index type on tsvector columns. The question asks for index type. In Cloud SQL for PostgreSQL, you can use the built-in full-text search with GIN indexes.

Answer

B-tree index

Answer

GiST index

Answer

Hash index

Question 13

A team wants to implement chaos engineering on GKE to test resilience. Which THREE fault types can be injected using Chaos Mesh? (Choose 3 answers)

Accepted Answer

Pod failure (kill pods).. Chaos Mesh is a CNCF chaos engineering platform for Kubernetes that provides a set of CRD-based fault injectors, and three of the listed fault types map directly to its built-in injectors. Option A (Pod failure / kill pods) is correct because Chaos Mesh's PodChaos supports actions like pod-kill and pod-failure to terminate or make pods unavailable, testing workload resilience and restart behavior. Option D (CPU stress) is correct because StressChaos injects CPU (and memory) stress into selected pods via stressors, simulating resource contention on GKE nodes. Option E (Network latency) is correct because NetworkChaos can add latency, packet loss, duplication, and corruption to pod traffic using tc/netem rules, validating service timeouts and retries. Option B (code injection into running containers) is not a Chaos Mesh fault type; it is not supported as an injector and would require runtime debugging tools rather than chaos CRDs. Option C (whole cluster deletion) is not offered by Chaos Mesh, which operates at pod, network, stress, IO, DNS, and time levels within a cluster, not by deleting the entire GKE cluster.

Answer

Code injection into running containers.

Answer

Whole cluster deletion.

Question 14

An organization wants to track cloud costs per team and per project. They have already enabled billing export to BigQuery. What additional step should they take to enable cost attribution?

Accepted Answer

Apply labels to resources (team, environment) and query BigQuery billing export. Cost attribution in GCP relies on labels applied to resources (e.g., team, environment, project) combined with the BigQuery billing export. Once labels exist, you can query the exported billing dataset to group and filter costs by label, enabling per-team and per-project chargeback.

Answer

Create separate billing accounts per team

Answer

Use resource hierarchy folders to track costs

Answer

Enable VPC flow logs for cost tracking

Question 15

A company uses Cloud Logging and wants to receive real-time notifications when the number of 5xx errors exceeds 100 per minute. They need to set up an alerting policy. Which metric type and condition should they use?

Accepted Answer

Create a log-based counter metric and use a metric-threshold condition with rate alignment. A log-based counter metric counts matching log entries, and a metric-threshold condition with rate alignment evaluates the count per unit time (e.g., per minute). This directly matches the requirement to alert when 5xx errors exceed 100 per minute, because rate alignment converts the counter into a per-minute rate for comparison against the threshold.

Answer

Create a log-based counter metric and use a metric-absent condition

Answer

Use a Cloud Monitoring gauge metric with a custom value

Answer

Use a logs-based alert with a threshold on the log entry count