Google Cloud Professional Cloud Database Engineer practice questions

Question 1

A platform team is designing a Memorystore for Redis cache with a replica for high availability and must set expectations about potential data loss if the primary fails. Which statement correctly describes how Memorystore for Redis replicates data to its replicas?

Accepted Answer

Memorystore for Redis asynchronously copies primary changes to the replica, so a small data-loss window is possible.

Answer

Memorystore for Redis replicates to the replica synchronously, so failover preserves every acknowledged write to the primary.

Answer

Memorystore for Redis does not replicate at all; the replica is an empty standby populated only after a failover happens.

Answer

Memorystore for Redis shards keys across the replicas without copying them, so each replica holds a distinct data subset.

Question 2

A team is being asked to grant Firestore access to a vendor service account that needs to read and write entities but must not be able to manage indexes or backups. The team is evaluating Firestore's predefined IAM roles. Which role assignment matches the requirement?

Accepted Answer

The platform offers predefined roles like datastore.owner for full access, datastore.user for read/write data access, and datastore.viewer for read-only access, and permissions are organized by category including database operations, entity management, index operations, and backup administration.

Answer

Assign datastore.owner to give the vendor unrestricted ownership of the database. Owner grants full access including index and backup management, which exceeds the requirement.

Answer

Assign roles/iam.serviceAccountUser only and rely on resource-level Firestore filtering. iam.serviceAccountUser is unrelated to Firestore database operations. This contradicts how the cited Google Cloud product guide describes the supported configuration, so a senior database engineer reviewing the design would flag it during architecture review rather than approve it.

Answer

Firestore does not support service-account access; only human users can be granted roles. The documentation explicitly describes service-account access patterns via IAM roles.

Question 3

An organization is being asked to assess its on-premises database and server estate before migrating to Google Cloud, including cost estimation, asset discovery, and dependency mapping. The team is evaluating Migration Center. Which description matches its documented role?

Accepted Answer

Google Cloud Migration Center is a unified platform helping organizations accelerate cloud journeys from on-premises or other cloud environments to Google Cloud, enables cost estimation, automated asset discovery of servers and databases, infrastructure assessment with dependency identification, migration planning with risk mitigation, and execution support across various migration strategies including rehost, replatform, and refactor approaches.

Answer

Migration Center is solely a Compute Engine sizing calculator and does not assess databases. The documentation explicitly lists automated asset discovery of servers and databases.

Answer

Migration Center is a fully managed database engine that hosts a temporary copy of the workload during migration. Migration Center is an assessment and planning platform, not a database engine.

Answer

Migration Center requires the team to disable Cloud Monitoring before it can assess the estate. The documentation does not require disabling Cloud Monitoring; this is fabricated. The Google Cloud documentation does not endorse this configuration for the stated workload, and choosing it here would introduce risk that the cited product guidance is explicitly written to avoid.

Question 4

A company is deploying a Spanner database for a globally distributed application and is evaluating whether a regional or a multi-region instance configuration better fits low-latency reads for users on several continents. Which statement matches the documentation?

Accepted Answer

A multi-region configuration spans resources across multiple regions, making it the better fit for a globally distributed user base.

Answer

A regional configuration spreads resources across multiple regions, so it best fits a user base spread across several continents.

Answer

The instance configuration affects only billing, not where data is stored, so either choice yields the same global latency profile.

Answer

Multi-region is unavailable for new instances, so the company must run several separate regional instances and keep them in sync itself.

Question 5

A platform team is designing a Cloud SQL for PostgreSQL HA configuration for an order-processing workload and must explain the regional-instance topology to auditors, including the role of the standby instance. The team is evaluating which characterization matches the documented HA model. Which description is accurate?

Accepted Answer

A Cloud SQL instance configured for HA is also called a regional instance and has a primary and secondary zone within the configured region, and within a regional instance, the configuration is made up of a primary instance and a standby instance.

Answer

The HA configuration runs two independent primaries in two regions and applications must implement client-side conflict resolution. Cloud SQL HA is in-region and not a multi-region active-active topology.

Answer

The HA configuration uses two read-only replicas in the same zone as the primary and rotates them via DNS during failover. Two co-zonal read replicas would not survive a zonal failure, so this is not the HA topology.

Answer

The HA configuration disables automatic failover and requires an SRE to run a manual command for any zone outage. Cloud SQL HA includes automatic failover; the standby exists precisely so failover can occur without operator action.

Question 6

A company needs to continuously capture row-level changes from heterogeneous MySQL, PostgreSQL, and Oracle source databases and stream them into BigQuery with little infrastructure to manage. The team must choose between Google Cloud data services. Which choice best fits?

Accepted Answer

Use Datastream, which represents the differing source types with a unified, lossless model across sources.

Answer

Use Dataflow templates, which are built for general batch and streaming processing rather than change-data capture.

Answer

Use BigQuery federated queries, since EXTERNAL_QUERY against each source removes any need to replicate the data.

Answer

Use external read replicas of each database, then run a nightly export of their contents into BigQuery tables.

Question 7

A cloud administrator has configured a Storage Transfer Service job to pull objects from an Amazon S3 bucket, but the AWS account enforces IP restrictions on storage access. To keep the agentless transfer working as Google's worker IPs change, what should the administrator do per the documentation?

Accepted Answer

Allowlist the published worker IP ranges and re-pull the JSON list weekly, as new ranges wait 7 days.

Answer

Allowlist one single static IP that Storage Transfer Service permanently reserves for every customer.

Answer

Disable all IP restrictions on the S3 bucket since Storage Transfer Service cannot work with allowlists.

Answer

Allowlist your own agent machine IPs, which is required for every agentless Storage Transfer job here.

Question 8

A company runs a latency-sensitive MySQL workload on Cloud SQL and is being squeezed by maintenance restarts; an engineer is tasked with cutting maintenance downtime to a minimum and confirming how hardware updates are handled. Which statement matches the documented behavior?

Accepted Answer

Cloud SQL Enterprise Plus edition can limit maintenance downtime to sub-second, and hardware updates use live migration with no interruption.

Answer

Switching to Cloud SQL Enterprise edition removes maintenance entirely, and hardware updates are skipped so the instance never restarts at all.

Answer

Maintenance downtime can only be avoided by self-managing MySQL on Compute Engine, since Cloud SQL cannot reduce restart downtime.

Answer

Hardware updates require a full instance restart, while Enterprise Plus reduces only the frequency, not the length, of maintenance.

Question 9

A company is migrating a MySQL application to Cloud SQL and must pick an edition for a workload that needs the best performance, availability, and observability. The team is evaluating how the Enterprise and Enterprise Plus editions differ. Which statement correctly distinguishes them?

Accepted Answer

Enterprise Plus provides the best performance, availability, and observability; Enterprise covers core needs.

Answer

Enterprise edition delivers the best performance and observability; Enterprise Plus only adds longer backups.

Answer

Both editions are identical in performance; the edition name only changes the monthly billing currency used.

Answer

Enterprise Plus is a NoSQL-only tier, while Enterprise edition is the sole option for relational MySQL data.

Question 10

A platform team is migrating away from shared database passwords on a Cloud SQL for PostgreSQL fleet and must centralize authentication on Google identities, including group membership. The team is evaluating Cloud SQL IAM database authentication. Which statement correctly describes its behavior?

Accepted Answer

IAM database authentication uses temporary access tokens valid for one hour and requires SSL connections for security, and the system supports both individual user authentication and group-based access management through Cloud Identity groups.

Answer

IAM database authentication issues long-lived 90-day passwords stored in Secret Manager and replaces username/password auth without requiring SSL. The documentation describes temporary one-hour access tokens and requires SSL connections.

Answer

IAM database authentication is limited to a single individual user per Cloud SQL instance and does not support group-based access management. The documentation explicitly supports group-based access management through Cloud Identity groups.

Answer

IAM database authentication forces the application to embed a static service-account JSON key in every connection string. The documentation describes short-lived access tokens, not static service-account JSON keys.

Question 11

A team is being tasked with migrating a MySQL source on Azure Database for MySQL into Cloud SQL using DMS. They are preparing the migration user account and reviewing transport requirements for the source. Which constraint and configuration combination matches the documentation?

Accepted Answer

The user password cannot exceed 32 characters, and for Azure sources, check the require_secure_transport setting to determine encryption requirements.

Answer

DMS imposes a 64-character password limit and does not consider any Azure-specific transport setting. The documentation explicitly caps the password at 32 characters and instructs administrators to check Azure's require_secure_transport setting.

Answer

DMS does not support Azure Database for MySQL as a source. The documentation explicitly lists Microsoft Azure Database for MySQL versions 5.5 through 8.4 as a supported source.

Answer

DMS forces the team to disable SSL on the Azure source. The documentation says require_secure_transport must be checked to determine encryption requirements, not that SSL must be disabled.

Question 12

A developer is deploying a Memorystore for Redis Cluster instance and needs to understand how a single shard is composed before sizing the cluster. Which statement matches the documented architecture?

Accepted Answer

Each shard has one primary node and optionally up to five replica nodes, and each shard holds a subset of the key space.

Answer

Each shard has exactly two primary nodes and no replicas, with the key space fully duplicated on every shard in the cluster.

Answer

Each shard is a single node that stores the entire key space, so adding shards only increases memory, not partitioning.

Answer

Each shard must contain at least three replica nodes before any primary node can accept writes from the application clients.

Question 13

An architect is designing a Spanner schema for a high-write event-tracking table and is being asked to prevent hotspots that could saturate a single server's resources. The team must select primary-key strategies that the Spanner schema-design documentation recommends. Which two strategies are documented as ways to distribute write load evenly across servers? (Choose two.)

Accepted Answer

One recommendation is using UUIDs, which function independently of the record's content and don't maintain locality between related records.

Accepted Answer

Other techniques include bit-reversed sequences, key reordering, or hash-based sharding to distribute data evenly across servers.

Answer

Use a monotonically increasing timestamp column as the first key part to keep recent rows physically co-located. The documentation explicitly says monotonically increasing or decreasing values such as a timestamp should be avoided as the first key part for high write rate tables.

Answer

Use a single fixed-value sharding column repeated for every row so the storage engine can amortize lookups against one shard. A single fixed value would concentrate writes on a single split and cause the hotspot the design is trying to avoid.

Question 14

An organization is querying a rarely accessed Bigtable table through a BigQuery external table and is seeing query latency slowly get worse over weeks of light use. An engineer is evaluating why performance degrades over time. Which documented behavior explains it?

Accepted Answer

Bigtable auto-merges splits based on load, so a rarely read table keeps fewer splits and queries gradually slow.

Answer

BigQuery silently moves the external table to Google Drive after periods of low use, which is slower to scan.

Answer

The Cloud SQL maintenance window restarts the Bigtable instance weekly, adding latency to every external query.

Answer

BigQuery caches fewer column families for idle tables, so each query re-reads the entire row from Colossus.

Question 15

A migration engineer is being tasked with preparing a self-managed MySQL 8.0 source database for Database Migration Service to Cloud SQL for MySQL. The engineer must enable binary logging and configure the source so DMS can perform the full dump and continuous replication. Which configuration is documented as required?

Accepted Answer

Before migrating, you must stop DDL write operations during the full dump phase, set the server-id option to 1 or larger, configure GTID logging by setting GTID_MODE to ON or OFF, enable binary logs with row-based logging, ensure all tables use InnoDB storage engine, and create a separate user account with appropriate privileges depending on your migration type.

Answer

Configure the source MySQL instance with the MyISAM storage engine for all tables to speed up the full dump phase. The documentation explicitly requires all tables to use the InnoDB storage engine.

Answer

Enable statement-based binary logging only; row-based logging is not supported. The documentation explicitly requires row-based binary logging. This approach diverges from the documented behavior and would not satisfy the design intent the team is being asked to deliver under the cited Google Cloud product guide.

Answer

Set server-id to 0 on every source so DMS can auto-assign IDs at runtime. The documentation explicitly requires server-id to be set to 1 or larger.

Google Cloud Professional Cloud Database Engineer

Sample questions

Sources

Similar exams