PCDE Practice Test 27 — 15 Questions

Question 1

A service has an SLO based on request latency: 99% of requests must complete under 500ms over a 28-day window. The team wants to monitor the error budget burn rate. Which Cloud Monitoring SLO type and configuration should be used?

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

Accepted Answer

Request-based SLO: good request count / valid request count with threshold 500ms. A request-based SLO counts individual requests and classifies them as good or bad based on a threshold — here, requests completing under 500ms are 'good.' The correct denominator is 'valid request count' (requests that are eligible for the SLO, excluding those that shouldn't count, like client errors or health checks). This matches Cloud Monitoring's request-based SLO model for latency.

Answer

Window-based SLO: good minutes / total minutes with threshold 500ms

Answer

Request-based SLO: good request count / total request count with threshold 500ms

Answer

Window-based SLO: good requests / total requests with threshold 500ms

Question 2

A financial services company uses Cloud Spanner with a database that has multiple tables with interleaved relationships. They need to enforce a strict consistency requirement across two related tables that are not interleaved. Which method ensures global strong consistency?

Accepted Answer

Use Spanner's built-in atomicity by executing the updates in a single read-write transaction.. Cloud Spanner provides external consistency (global strong consistency) across all tables, interleaved or not, through the use of distributed read-write transactions that leverage the TrueTime API. By executing updates to both non-interleaved tables within a single read-write transaction, Spanner ensures that all mutations are applied atomically and are visible globally at a single timestamp, meeting the strict consistency requirement.

Answer

Use Cloud Pub/Sub to eventually synchronize the tables.

Answer

Use a commit timestamp-based approach to synchronize writes.

Answer

Use a client-side distributed transaction across the two tables.

Question 3

A developer needs to connect to a Memorystore for Redis instance from a Compute Engine VM in the same VPC network. The Redis instance has AUTH enabled. What must the developer provide in the connection string?

Accepted Answer

The Redis AUTH password.. When AUTH is enabled on a Memorystore for Redis instance, the client must provide the AUTH password in the connection string (e.g., `redis://:password@host:6379`) or via the `AUTH` command after connecting. This password is set during instance creation and is required for authentication before any data commands can be executed. Without it, the Redis server will reject the connection with a `NOAUTH Authentication required` error.

Answer

The IAM role 'Memorystore User'.

Answer

The client certificate and private key.

Answer

The service account key file.

Question 4

Which TWO actions should be taken to enable layer caching for Docker builds using Kaniko in Cloud Build? (Choose two.)

Accepted Answer

Add `--cache-repo` specifying an Artifact Registry repository. Option D is correct because Kaniko only enables layer caching when the `--cache=true` flag is explicitly passed to the executor; without it, Kaniko rebuilds every layer regardless of any repository configuration. Option A is correct because `--cache-repo` tells Kaniko where to store and retrieve the cached layers (an Artifact Registry repository), which is required for the cache to persist across builds; `--cache=true` alone has nowhere to read/write cache data. Option B is wrong because switching to the `docker` builder abandons Kaniko entirely and does not configure Kaniko caching. Option C is wrong because `CACHE_LAYERS` is not a Cloud Build substitution variable that Kaniko recognizes. Option E is wrong because Cloud Build has no `cache: true` build-step field that enables Kaniko layer caching.

Answer

Use the `docker` builder instead of Kaniko

Answer

Set the `CACHE_LAYERS` substitution variable

Answer

Enable Cloud Build's built-in caching by adding `cache: true`

Question 5

A company runs a Cloud Bigtable instance with a single cluster in us-central1. They want to implement a disaster recovery plan that meets an RPO of 5 minutes and an RTO of 10 minutes in case of a regional outage. Which actions should they take?

Accepted Answer

Add a second cluster in a different region, configure replication, and use multi-cluster routing. Cloud Bigtable replication requires a second cluster in a different region within the same instance; multi-cluster routing automatically routes reads and writes to the nearest available cluster, enabling failover with replication lag typically under a few seconds — well within the 5-minute RPO. RTO is near-instant because the second cluster is already running and serving traffic. This is the only option that provides both multi-region replication and automatic failover.

Answer

Use Bigtable replication with 'read-failover' routing policy and use Cloud Functions to failover

Answer

Use Cloud SQL for MySQL with cross-region replica for better RPO

Answer

Create a scheduled backup every 5 minutes to Cloud Storage and restore in another region on failure

Question 6

A company has a BigQuery table partitioned by ingestion time. They want to create a BI report showing month-over-month revenue growth. To minimize query cost, what should they do?

Accepted Answer

Use a WHERE clause with _PARTITIONDATE >= DATE_SUB(CURRENT_DATE(), INTERVAL 13 MONTH) and LAG. It uses a WHERE clause with _PARTITIONDATE >= DATE_SUB(CURRENT_DATE(), INTERVAL 13 MONTH) to prune partitions, ensuring BigQuery scans only the necessary 13 months of data. The LAG function then computes month-over-month revenue growth efficiently. This minimizes query cost by reducing the amount of data processed, which is critical for ingestion-time partitioned tables.

Answer

Use DATE_TRUNC on the ingestion timestamp without filtering partitions

Answer

Use LAG without a partition filter

Answer

Use a wildcard table with UNION ALL over monthly tables

Question 7

An engineer is manually migrating a MySQL database to Cloud SQL using mysqldump and import. They need to capture the binary log position to enable CDC with Database Migration Service later. Which TWO mysqldump flags should they include to ensure a consistent snapshot and capture the log position? (Choose 2 correct answers.)

Accepted Answer

--single-transaction. Option B (--single-transaction) is correct because it makes mysqldump issue a START TRANSACTION WITH CONSISTENT SNAPSHOT, giving a consistent InnoDB snapshot without locking tables, which is essential for a clean migration baseline. Option C (--master-data=2) is correct because it writes the binary log file and position into the dump as a comment (value 2 = commented, non-executable), which is exactly the log position needed to later configure CDC in Database Migration Service. Option A (--no-data) is wrong because it would dump only the schema and omit the data needed for the initial snapshot. Option D (--skip-lock-tables) is wrong because it disables table locking without providing snapshot consistency or log-position capture. Option E (--triggers) is wrong because it only includes trigger definitions in the dump and has nothing to do with consistent snapshots or binary log positions.

Answer

--no-data

Answer

--skip-lock-tables

Answer

--triggers

Question 8

A company wants to use Cloud Logging to monitor for a specific error pattern in their application logs. They want to create a metric that counts occurrences of the error and then set up an alert when the count exceeds 100 in 5 minutes. Which TWO components are required? (Choose 2)

Accepted Answer

Alerting policy with metric threshold condition. Option E (Log-based counter metric) is correct because counting occurrences of a specific error pattern in logs requires a log-based metric of type counter, which increments by 1 for each matching log entry and exposes that count to Cloud Monitoring. Option C (Alerting policy with metric threshold condition) is correct because an alert fires only when a Cloud Monitoring alerting policy evaluates the log-based counter metric against a threshold condition, here exceeding 100 over a 5-minute alignment period. Option A is not required because a notification channel is only needed to deliver the alert notification, not to create the metric or the alerting condition itself. Option B is incorrect because a distribution metric records value distributions (e.g., latencies) rather than counting occurrences. Option D is incorrect because a dashboard only visualizes metrics and plays no role in generating the alert.

Answer

Notification channel (e.g., email)

Answer

Log-based distribution metric

Answer

Cloud Monitoring dashboard

Question 9

An SRE team wants to implement chaos engineering on their GKE cluster. Which TWO options are valid tools or services for injecting faults into GKE workloads?

Accepted Answer

Chaos Mesh. Chaos Mesh (B) is a CNCF chaos-engineering platform that runs on Kubernetes and provides custom resources such as PodChaos, NetworkChaos, and StressChaos to inject faults like pod kills, network latency, and CPU stress directly into GKE workloads, making it a valid fault-injection tool. Traffic Director with HTTP fault filter (E) is also valid because Traffic Director, Google Cloud's managed service-mesh control plane, supports Envoy-based HTTP fault injection (abort and delay filters) that can inject HTTP errors or latency into traffic routed to GKE services. Cloud Audit Logs (A) only records administrative and data-access activity for auditing, not fault injection. Cloud Endpoints (C) is an API management and authentication layer, and Cloud Scheduler (D) is a cron-style job trigger; neither injects faults into workloads.

Answer

Cloud Audit Logs

Answer

Cloud Endpoints

Answer

Cloud Scheduler

Question 10

Refer to the exhibit. A data engineer created a materialized view on a table that receives streaming inserts. When they query the materialized view, they get this error. What is the most likely cause?

Accepted Answer

The materialized view cannot read data from the streaming buffer.. The error occurs because materialized views in BigQuery cannot directly read data from the streaming buffer. When a base table receives streaming inserts, the data resides in the streaming buffer for up to 90 minutes before being committed to storage. Materialized views only reflect committed data, so querying them during this window returns an error indicating that the view cannot access the streaming buffer.

Answer

The materialized view definition includes a JOIN that is not supported.

Answer

The materialized view has reached its maximum size limit.

Answer

The base table has a schema change that the materialized view cannot adapt to.

Question 11

A company is migrating a large Oracle Data Warehouse to BigQuery. The source schema includes many partitioned tables and materialized views. Which THREE considerations are important when designing the BigQuery schema?

Accepted Answer

Clustering can be used to improve query performance on frequently filtered columns.. Option A is correct because BigQuery clustering sorts data within partitions by up to four columns, so filtering or aggregating on frequently used columns (e.g., customer_id) reduces bytes scanned and improves performance. Option B is correct because BigQuery supports partitioning by a DATE or TIMESTAMP column, by an INTEGER column (e.g., range partitioning on a numeric key), and also by ingestion time, which is essential when mapping Oracle partitioned tables. Option D is correct because BigQuery materialized views are automatically refreshed in the background when base table data changes, with refresh intervals configurable, so migrated Oracle materialized views do not need manual refresh jobs. Option C is not correct because BigQuery is a columnar, serverless engine that does not require or support user-created indexes on WHERE-clause columns; it relies on partitioning, clustering, and automatic storage optimizations instead. Option E is not correct because BigQuery only enforces NOT NULL and does not enforce unique constraints or foreign keys (they can be declared for informational purposes but are not enforced), so data integrity must be handled by the ETL process.

Answer

BigQuery requires explicit indexes on columns used in WHERE clauses.

Answer

BigQuery supports unique constraints and foreign keys for data integrity.

Question 12

An organization wants to enforce security policies on container images before deployment. They need to scan images for vulnerabilities and ensure only images that pass the scan can be deployed. Which TWO services should they use? (Select TWO)

Accepted Answer

Container Analysis. Container Analysis (A) is correct because it is the Google Cloud service that performs vulnerability scanning on container images stored in Artifact Registry or Container Registry, producing metadata about known CVEs in the image layers. Binary Authorization (E) is correct because it enforces deploy-time admission control on Google Kubernetes Engine and Cloud Run, allowing only images that satisfy attestations or have passed required scans to be deployed. Together they satisfy the requirement to scan images and block non-compliant ones. Artifact Registry (B) only stores and manages container images and does not itself enforce deployment policy. Cloud Build (C) is a CI/CD build service that can run scanners but is not the policy enforcement mechanism. Cloud Deploy (D) is a delivery pipeline service for releasing to GKE and does not provide image vulnerability scanning or admission control.

Answer

Artifact Registry

Answer

Cloud Build

Answer

Cloud Deploy

Question 13

A data engineer creates a clustered table in BigQuery with clustering order: country, city, product_id. The BI team frequently runs a query that filters on city and product_id but rarely on country. What is the most likely performance issue?

Accepted Answer

The query does not filter on the first clustering column (country), so block pruning is minimal.. BigQuery clustered tables use block pruning to skip reading blocks that don't match the query's filter. Pruning is most effective when the filter includes the first clustering column (country). Without it, BigQuery must scan more blocks, leading to higher query costs and slower performance.

Answer

BigQuery allows only one clustering column per table.

Answer

The table should be partitioned by country instead of clustered.

Answer

The query filters on too many clustering columns, causing overhead.

Question 14

A team is migrating to GCP and wants to use Cloud Deployment Manager for infrastructure. They have existing Terraform modules. What is the best approach?

Accepted Answer

Use Terraform with a GCS backend; avoid Deployment Manager.. The team already has existing Terraform modules, and using Terraform with a GCS backend allows them to continue leveraging their investment in Terraform without rewriting or bridging tools. Cloud Deployment Manager is a native Google Cloud IaC tool, but it does not natively import or execute Terraform configurations; forcing a migration or hybrid approach introduces unnecessary complexity and risk. The best practice is to standardize on one IaC tool, and since Terraform is already in use, using it with a GCS backend for state management is the most efficient and reliable path.

Answer

Use Deployment Manager only for new resources and Terraform for existing.

Answer

Convert Terraform modules to Deployment Manager templates manually.

Answer

Run Terraform as a subprocess within Deployment Manager.

Question 15

During a postmortem, an SRE team identifies several contributing factors. Which THREE items should be included in the action items section of a blameless postmortem?

Accepted Answer

Verification steps to confirm the fix is effective. Option E is correct because effective action items must be concrete, specific, and tied to the identified root causes, with exactly one accountable owner per item so responsibility is unambiguous. Option B is correct because each action item should carry a due date, which makes the remediation trackable and prevents items from lingering indefinitely after the postmortem. Option A is correct because including verification steps ensures the fix is actually validated as effective rather than assumed, closing the loop on the remediation. Option C is not appropriate because vague, generic recommendations such as 'improve testing' are not actionable or measurable. Option D is not appropriate because blameless postmortems focus on systemic causes and process improvements, not on assigning blame to individuals.

Answer

Generic recommendations like 'improve testing'

Answer

Assign blame to the engineer who caused the incident