Free Professional Cloud Database Engineer Practice Questions

Sample Practice Questions

Try these 10 sample questions from the Professional Cloud Database Engineer practice bank. Each includes the answer and explanation. Continue with the interactive practice above to review the full question set.

1. 1When sizing memory for a performance-sensitive Cloud SQL for PostgreSQL OLTP workload, which approach follows the documented guidance?

   1. A. Size memory from total database size and omit connection overhead.
   2. B. Fit the working set in memory, account for connections, and load test.
   3. C. Size memory from vCPU count alone and validate only disk capacity.
   4. D. Fit only index metadata in memory and size for idle connection counts.

   Answer: B

   Explanation: The guidance says, "enough memory to contain the entire working set" and identifies "number of active connections" as another factor. It also says, "Perform load testing to avoid performance issues". Sizing must therefore include the working set and connection demand and be validated under load.

   Reference: Cloud SQL for PostgreSQL instance settings — Memory

2. 2An IAM principal has been added as a Cloud SQL database user but lacks the IAM permission required for database login. Which predefined role specifically supplies cloudsql.instances.login?

   1. A. Cloud SQL Viewer (roles/cloudsql.viewer).
   2. B. Cloud SQL Instance User (roles/cloudsql.instanceUser).
   3. C. Cloud SQL Client (roles/cloudsql.client).
   4. D. Cloud SQL Schema Viewer (roles/cloudsql.schemaViewer).

   Answer: B

   Explanation: The login guide identifies "roles/cloudsql.instanceUser" as containing "cloudsql.instances.login". The official IAM reference confirms that permission in the Instance User role. Among these options, Cloud SQL Viewer, Client, and Schema Viewer do not include the database-login permission.

   Reference: Log in using IAM database authentication — Before you begin — IAM roles

3. 3A continuous PostgreSQL migration to Cloud SQL is ready for cutover. The priority is to avoid omitting outstanding source changes. What sequence follows the promotion guidance?

   1. A. Promote first, keep source writers running, and wait for lag to decrease afterward.
   2. B. Redirect writes to both databases, promote, and reconcile conflicts after cutover.
   3. C. Stop source writes and clients, wait for near-zero replication delay, then promote.
   4. D. Stop destination replication immediately and redirect clients before source writes stop.

   Answer: C

   Explanation: The guide says, "Stop all writes, running scripts, and client connections to the source database." It then says to wait until replication delay is at or close to zero. Promoting with outstanding lag may affect destination data accuracy, so continuing source writes or disconnecting replication first contradicts the stated priority.

   Reference: Promote a PostgreSQL migration job — Promotion procedure — stop writes and check replication delay

4. 4An engineer is provisioning a new Cloud SQL for PostgreSQL instance with gcloud. Which availability setting creates the instance with regional high availability?

   1. A. --availability-type=ZONAL
   2. B. --availability-type=REGIONAL
   3. C. --region=us-central1 without an availability-type setting
   4. D. --zone=us-central1-a without an availability-type setting

   Answer: B

   Explanation: The HA provisioning guide shows "gcloud sql instances create REGIONAL_INSTANCE_NAME" with "--availability-type=REGIONAL". A geographic region or zone selection alone does not request the regional HA configuration.

   Reference: Enable and disable high availability on an instance — Create a new instance configured for high availability — gcloud

5. 5A production Cloud SQL for PostgreSQL deployment needs an SLA-covered machine type. How should a proposal to use db-f1-micro or db-g1-small be assessed?

   1. A. Accept either type when automatic storage increases are enabled.
   2. B. Accept either type when the database uses a private IP address.
   3. C. Reject these types because shared-core types lack Cloud SQL SLA coverage.
   4. D. Reject db-f1-micro but accept db-g1-small for SLA-covered production.

   Answer: C

   Explanation: The documentation states, "The db-f1-micro and db-g1-small machine types aren't included in the Cloud SQL SLA." Both are shared-core test/development types. Private networking or automatic storage growth does not change their machine-type exclusion.

   Reference: Cloud SQL for PostgreSQL instance settings — Machine Type

6. 6An IAM database user can connect to Cloud SQL for PostgreSQL 16 and already has USAGE on schema public. It has no table privileges and must read rows from public.orders. Which table privilege should an administrator grant?

   1. A. INSERT on public.orders.
   2. B. UPDATE on public.orders.
   3. C. REFERENCES on public.orders.
   4. D. SELECT on public.orders.

   Answer: D

   Explanation: The IAM login guide says, "the new database user has no privileges to any database." PostgreSQL defines SELECT as: "Allows SELECT from any column, or specific column(s), of a table, view, materialized view, or other table-like object." With connection and schema access already supplied, granting SELECT on public.orders supplies the missing read privilege. INSERT, UPDATE, and REFERENCES do not grant ordinary SELECT access.

   Reference: Log in using IAM database authentication — Before you begin — database privileges

7. 7After a PostgreSQL DMS migration is promoted, the application writes new data to Cloud SQL. A fallback proposal simply points clients back to the original source, assuming DMS has copied those new writes backward. What is the flaw?

   1. A. Promotion disconnects the databases; source fallback needs a separate data-synchronization plan.
   2. B. Promotion keeps reverse replication active, but only database users require synchronization.
   3. C. Promotion continues bidirectional replication, but clients must wait for the next backup window.
   4. D. Promotion retains source-to-destination replication and requires reversing that live link for fallback.

   Answer: A

   Explanation: The guide says, "Promoting the migration job disconnects the destination instance from the source," and makes it independent. Consequently, promotion does not provide automatic reverse replication of subsequent destination writes. Returning to an unchanged old source can lose those writes unless the fallback plan synchronizes them separately.

   Reference: Promote a PostgreSQL migration job — Promotion overview

8. 8A team plans a manual Cloud SQL HA failover drill. The primary is operating normally, but the standby is unresponsive. What should it expect?

   1. A. Failover can proceed whenever the primary is healthy, regardless of standby health.
   2. B. Failover can proceed if the secondary zone exists, even with an unresponsive standby.
   3. C. Failover is blocked until the standby and secondary zone are healthy.
   4. D. Failover can proceed after a client reconnects successfully to the current primary.

   Answer: C

   Explanation: The HA requirements state, "When the standby instance is unresponsive, failover operations are blocked." Both the secondary zone and standby must be healthy. A manual drill does not bypass those eligibility requirements.

   Reference: About high availability (HA) — Failover overview — Requirements

9. 9A large import will rapidly consume free disk space on a Cloud SQL for PostgreSQL instance. Automatic storage increases are enabled. What should the capacity plan do before the import?

   1. A. Manually provision enough disk space for the expected import growth.
   2. B. Lower the storage increase limit to trigger growth more frequently.
   3. C. Leave disk allocation unchanged because automatic growth keeps pace.
   4. D. Disable automatic storage increases before starting the large import.

   Answer: A

   Explanation: The documentation says, "Before an operation that rapidly grows disk space usage," and specifically recommends "manually resize the disk, rather than" relying on automatic growth. Planning capacity before the import avoids assuming that reactive increments can keep up.

   Reference: Cloud SQL for PostgreSQL instance settings — Storage capacity — Considerations for large disks

10. 10A long-running service uses Cloud SQL IAM database authentication without context-aware access. New connections fail after it reuses a manually captured token for more than an hour. Which configuration addresses token renewal?

    1. A. Use automatic IAM database authentication through a supported Cloud SQL connector.
    2. B. Store the captured IAM token as the permanent PostgreSQL password for the user.
    3. C. Increase connection timeout so the expired token can finish authenticating.
    4. D. Disable SSL so IAM tokens can be accepted without an expiration check.

    Answer: A

    Explanation: IAM authentication tokens are "valid for only one hour." The documentation adds: "Cloud SQL connectors are able to request and refresh these tokens, ensuring that long-lived processes or applications that rely on connection pooling can have stable connections." Automatic IAM authentication through a supported connector handles renewal. Reusing the expired token, increasing a timeout, or disabling SSL does not renew it.

    Reference: IAM database authentication — Automatic versus manual IAM database authentication