I85PREP
⌘ K
GCP Database Engineer study materials
GCP Database Engineer study resources

GCP Database Engineer practice questions

One question at a time. Build your GCP Database Engineer understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Design scalable and highly available cloud database solutions

A team runs a Cloud SQL for PostgreSQL instance that must survive a zonal failure with automatic failover and no data loss. Which configuration should they choose?

  1. Zonal availability with automated backups enabled
  2. Regional availability with a standby in a second zone Correct answer.
  3. Single-zone instance with read replicas in other zones
  4. Zonal availability with point-in-time recovery enabled

Explanation

Cloud SQL regional (high availability) mode maintains a synchronous standby in a different zone; on failure the standby is promoted automatically and committed transactions are preserved. Zonal configurations and read replicas rely on asynchronous replication, so some committed data can be lost on failover.

Question 2Design scalable and highly available cloud database solutions

Which Spanner configuration provides the lowest write latency for a workload whose users are all in one metropolitan area?

  1. A regional configuration in that area Correct answer.
  2. A multi-region configuration with a single write region
  3. A dual-region configuration with one read-only region
  4. A multi-region configuration spanning continents

Explanation

A Spanner regional configuration keeps all replicas within one geographic area, so the leader and quorum are close to the application and commit latency stays low. Multi-region placements add wide-area round trips to the commit quorum.

Question 3Design scalable and highly available cloud database solutions

An application needs a globally distributed, strongly consistent relational database with horizontal write scaling. Which Google Cloud service is the intended fit?

  1. Firestore in Datastore mode
  2. Bigtable
  3. Cloud SQL for MySQL
  4. Spanner Correct answer.

Explanation

Spanner is Google Cloud's globally distributed relational database that offers external consistency and scales writes horizontally through sharding across nodes. The other options either lack relational semantics or do not scale writes horizontally in the same way.

Question 4Design scalable and highly available cloud database solutions

Which Bigtable schema design is recommended for a time-series workload that must serve recent data quickly and support range scans by entity?

  1. Row key of the timestamp alone, shared across all entities
  2. Row key of a random UUID followed by the timestamp
  3. Row key of entity ID followed by a reversed or zero-padded timestamp Correct answer.
  4. Row key of reversed timestamp followed by entity ID

Explanation

Prefixing the row key with the entity ID groups an entity's data into contiguous ranges for efficient scans, and appending a sortable timestamp keeps versions ordered within that range. This avoids hot-spotting because reads and writes spread across many entity prefixes.

Question 5Design scalable and highly available cloud database solutions

A Cloud SQL for PostgreSQL primary is under heavy read load and the team wants to offload analytics queries. What should they add?

  1. A larger machine tier for the primary
  2. Read replicas and direct analytics queries to them Correct answer.
  3. Binary logging with point-in-time recovery
  4. Cross-region automated backups

Explanation

Read replicas serve read-only traffic from a separate instance, removing analytical query load from the primary. Resizing the primary or enabling backups does not separate read traffic from the write path.

Question 6Design scalable and highly available cloud database solutions

Which AlloyDB for PostgreSQL feature lets a read pool scale independently of the primary without managing replica instances?

  1. Columnar engine for analytics acceleration
  2. Cross-region automated backups
  3. Read pools that add nodes to the instance Correct answer.
  4. Continuous backup and point-in-time recovery

Explanation

AlloyDB read pools add read nodes to an instance and scale read capacity independently of the primary. Backups and the columnar engine address recovery and query acceleration rather than read scaling.

Question 7Design scalable and highly available cloud database solutions

A Cloud SQL instance must be protected against a regional outage while keeping the primary's failover behavior. Which approach meets this requirement?

  1. Rely on the regional HA standby in the second zone
  2. Enable a cross-region read replica and promote it during a regional outage Correct answer.
  3. Increase the disk size to improve throughput
  4. Enable deletion protection on the instance

Explanation

A regional HA standby protects against a single zone failure but not a full regional outage. A cross-region read replica provides a copy in a different region that can be promoted during regional disaster recovery.

Question 8Design scalable and highly available cloud database solutions

Which Firestore mode should be selected for an application that needs strong consistency for transactions and simple queries but not SQL?

  1. Firestore in Native mode Correct answer.
  2. Spanner with a regional configuration
  3. Firestore in Datastore mode
  4. Bigtable with a single cluster

Explanation

Firestore in Native mode provides strong consistency for transactions and document queries with mobile and web SDK support. Datastore mode targets legacy App Engine workloads, while Bigtable and Spanner are different data models.

Question 9Design scalable and highly available cloud database solutions

A Bigtable cluster must remain available if one zone fails. What should be configured?

  1. An instance with replication disabled
  2. A single cluster with SSD storage
  3. A single cluster with one node
  4. A multi-cluster instance with clusters in separate zones Correct answer.

Explanation

A Bigtable instance with clusters in multiple zones replicates data across those zones so the instance remains available when one zone fails. A single-cluster instance has no redundancy across zones.

Question 10Design scalable and highly available cloud database solutions

Which Spanner feature supports reads from a nearby replica without sending the query to the leader?

  1. Interleaved tables
  2. Foreign keys with cascading deletes
  3. Stale reads using a timestamp bound Correct answer.
  4. Read-write transactions with pessimistic locking

Explanation

Spanner stale reads use a timestamp bound to read from a local replica at a consistent past point, reducing latency and leader load. Read-write transactions and schema features do not provide this locality benefit.

Question 11Design scalable and highly available cloud database solutions

Which design keeps Cloud SQL for MySQL writes available during a zone failure with minimal operational effort?

  1. Schedule frequent on-demand backups
  2. Attach a cross-region read replica
  3. Configure the instance as regional with automatic failover Correct answer.
  4. Use a larger machine type

Explanation

Regional Cloud SQL instances maintain a synchronous standby in another zone and fail over automatically. Backups and read replicas require manual recovery or promotion and do not provide automatic failover.

Question 12Design scalable and highly available cloud database solutions

A read-heavy Spanner workload is bottlenecked on CPU in the leader region. Which change most directly adds read capacity?

  1. Convert tables to interleaved tables
  2. Add more Spanner nodes to the instance Correct answer.
  3. Increase the timestamp bound for stale reads
  4. Enable the query optimizer version

Explanation

Spanner nodes provide compute and storage capacity; adding nodes increases the throughput available for reads and writes. The other options change query semantics or schema layout but do not add capacity.

Question 13Design scalable and highly available cloud database solutions

An application writes to Bigtable with row keys that are sequential numeric IDs, and writes concentrate on one tablet. What is the best fix?

  1. Hash or reverse the ID prefix to spread writes across tablets Correct answer.
  2. Increase the number of column families
  3. Move the table to SSD storage
  4. Enable a single-cluster instance

Explanation

Bigtable sorts rows lexicographically, so sequential keys create a hot spot at the end of the key space. Hashing or reversing the prefix distributes writes across tablets and improves throughput.

Question 14Design scalable and highly available cloud database solutions

Which Google Cloud database is best suited for a petabyte-scale analytics workload that requires very low-latency point lookups by row key rather than SQL joins?

  1. AlloyDB for PostgreSQL
  2. Bigtable Correct answer.
  3. Cloud SQL for SQL Server
  4. Spanner

Explanation

Bigtable is designed for very large volumes of data with low-latency reads and writes keyed by row, and it scales horizontally without SQL join semantics. The relational options are not intended for petabyte-scale key-value access.

Question 15Design scalable and highly available cloud database solutions

A Cloud SQL for PostgreSQL instance needs to fail over automatically when the primary zone becomes unavailable, and the team also wants to minimize failover time. Which configuration supports this?

  1. Zonal instance with a read replica in the same zone
  2. Regional instance with a standby in another zone Correct answer.
  3. Zonal instance with daily backups
  4. Zonal instance with binary logging enabled

Explanation

Regional instances maintain a synchronous standby in a different zone and promote it automatically when the primary fails. Zonal configurations lack automatic failover, and a same-zone replica shares the failure domain.

Question 16Design scalable and highly available cloud database solutions

Which AlloyDB feature accelerates analytical queries on transactional data by using an in-memory columnar format?

  1. Continuous backup
  2. Columnar engine Correct answer.
  3. Private Service Connect
  4. Read pools

Explanation

The AlloyDB columnar engine maintains an in-memory columnar representation of table data so analytical scans run faster alongside transactional workloads. Read pools, backups, and networking features address different concerns.

Question 17Design scalable and highly available cloud database solutions

Which practice improves the availability of a Spanner-backed application during a regional outage?

  1. Store all data in interleaved tables
  2. Use a regional instance with a single node
  3. Use read-write transactions exclusively
  4. Use a multi-region instance configuration Correct answer.

Explanation

A Spanner multi-region configuration keeps replicas in multiple regions so the database remains available when one region fails. A regional instance, interleaving, and transaction mode do not add geographic redundancy.

Question 18Design scalable and highly available cloud database solutions

A Cloud SQL for MySQL read replica is lagging significantly behind the primary. Which action best restores acceptable read freshness?

  1. Shorten the backup retention window
  2. Enable deletion protection
  3. Reduce write load on the primary and scale the replica's machine tier Correct answer.
  4. Increase the primary's disk size

Explanation

Replica lag usually reflects insufficient apply capacity or excessive write volume. Lowering write pressure and giving the replica more CPU and memory lets it apply the replication stream faster. Disk size, deletion protection, and backup retention do not affect replication throughput.

Question 19Design scalable and highly available cloud database solutions

Which Firestore configuration provides the strongest consistency guarantee for multi-document transactions?

  1. Firestore in Native mode with a single-region location and no transactions
  2. Firestore in Native mode with multi-region location Correct answer.
  3. Firestore in Datastore mode with eventual consistency
  4. Firestore in Datastore mode with a multi-region location

Explanation

Firestore in Native mode supports strongly consistent multi-document transactions, and multi-region locations replicate data while preserving those guarantees. Datastore mode and non-transactional configurations do not offer the same semantics.

Question 20Design scalable and highly available cloud database solutions

A team wants a managed PostgreSQL-compatible database that scales read traffic with additional nodes and offers a columnar engine for analytics. Which service fits?

  1. Bigtable
  2. Cloud SQL for PostgreSQL
  3. Firestore in Native mode
  4. AlloyDB for PostgreSQL Correct answer.

Explanation

AlloyDB for PostgreSQL combines PostgreSQL compatibility with read pools for scaling reads and a columnar engine for analytical acceleration. Cloud SQL lacks the columnar engine, and the other options are not PostgreSQL-compatible relational databases.

Question 21Design scalable and highly available cloud database solutions

An application performs many small transactions against Spanner in a single region and needs the lowest possible commit latency while retaining multi-zone durability. Which configuration is most appropriate?

  1. A multi-region configuration with three continents
  2. A regional configuration with replicas across zones in that region Correct answer.
  3. A multi-region configuration with two read-write regions
  4. A regional configuration with a single replica

Explanation

A regional Spanner configuration places replicas across zones within one region, giving multi-zone durability with commit quorums that stay local. Multi-region configurations add cross-region round trips that increase commit latency.

Question 22Design scalable and highly available cloud database solutions

Which Bigtable setting should be used to keep an instance available across zones while allowing reads to be served from the closest cluster?

  1. Multi-cluster routing with clusters in multiple zones Correct answer.
  2. Single-cluster instance with SSD storage
  3. Single-cluster instance with HDD storage
  4. An instance with a single node and no replication

Explanation

Multi-cluster routing replicates data across clusters in different zones and lets the client route reads to a nearby cluster. Single-cluster configurations offer neither zone redundancy nor read locality.

Question 23Design scalable and highly available cloud database solutions

A Cloud SQL for PostgreSQL instance must support both high availability and read scaling for reporting. Which combination meets both requirements?

  1. Regional configuration plus read replicas Correct answer.
  2. Single-zone instance plus a larger disk
  3. Zonal configuration plus point-in-time recovery
  4. Zonal configuration plus automated backups

Explanation

A regional configuration supplies a synchronous standby for automatic failover, while read replicas add separate instances for reporting traffic. Backups, point-in-time recovery, and larger disks address recovery and storage rather than availability and read scaling.

Question 24Design scalable and highly available cloud database solutions

A team needs a globally distributed database that supports SQL and external consistency for financial ledgers across continents. Which service is designed for this requirement?

  1. Bigtable with multi-cluster routing
  2. Firestore in Datastore mode
  3. Spanner with a multi-region configuration Correct answer.
  4. Cloud SQL for PostgreSQL with cross-region replicas

Explanation

Spanner multi-region configurations provide globally distributed replicas with external consistency and SQL support, which suits cross-continent financial ledgers. The alternatives either lack global strong consistency or relational SQL semantics.

Question 25Design scalable and highly available cloud database solutions

Which AlloyDB capability allows a read pool to serve queries while the primary handles writes, and how does it differ from Cloud SQL read replicas?

  1. Read pools store data in a columnar format for analytics
  2. Read pools add nodes to the instance and scale reads independently Correct answer.
  3. Read pools replace the primary during failover
  4. Read pools replicate data asynchronously to another region

Explanation

AlloyDB read pools attach additional read nodes to the same instance, scaling read capacity independently of the primary. Cloud SQL read replicas are separate instances with their own endpoints.

Question 26Design scalable and highly available cloud database solutions

A team is choosing between Cloud SQL and Spanner for a new workload that must survive the loss of an entire region while continuing to accept writes. Which property of Spanner is decisive for this requirement?

  1. Connection pooling through the Cloud SQL Auth Proxy
  2. Synchronous multi-region replication with quorum commit across regions Correct answer.
  3. Native export to Cloud Storage for backups
  4. Automated storage growth with no capacity planning

Explanation

Spanner replicates data synchronously across regions and commits using a quorum, so a multi-region configuration keeps accepting writes when one region is lost. Cloud SQL regional availability protects against a zone failure but not a full regional outage with automatic write continuity.

Question 27Design scalable and highly available cloud database solutions

An operations team wants Cloud SQL for PostgreSQL to survive a zonal failure without manual intervention. Which configuration satisfies this with the least operational effort?

  1. A cross-region read replica promoted by script
  2. Point-in-time recovery enabled on the primary
  3. A regional instance with a standby in a second zone Correct answer.
  4. A single-zone instance with automated daily backups

Explanation

A regional Cloud SQL instance maintains a standby in a different zone and fails over automatically when the primary zone is lost. Backups and point-in-time recovery restore data after an event but do not provide automatic failover.

Question 28Design scalable and highly available cloud database solutions

A Bigtable instance must remain readable and writable when a single zone becomes unavailable. Which instance configuration should be selected?

  1. A single-cluster instance in one zone
  2. A single cluster with a higher node count
  3. An instance using SSD storage in one zone
  4. A multi-cluster instance with clusters in separate zones Correct answer.

Explanation

Bigtable achieves zone resilience by placing clusters in different zones within a multi-cluster instance, so replication keeps data available when one zone fails. Adding nodes or changing storage type within one zone does not protect against zonal loss.

Question 29Design scalable and highly available cloud database solutions

Which Firestore mode should be chosen for a mobile application that needs multi-document transactions with strong consistency and does not require SQL?

  1. Datastore mode with eventual consistency
  2. Native mode with multi-region location Correct answer.
  3. Native mode in a single region without transactions
  4. Datastore mode restricted to a single zone

Explanation

Firestore Native mode supports ACID transactions across documents and can be deployed in a multi-region location for availability. Datastore mode is intended for server-side Datastore API workloads and does not offer the same Native mode client feature set.

Question 30Design scalable and highly available cloud database solutions

A Cloud SQL for MySQL primary in one zone must keep writes available through a zone failure with minimal operational effort. What should be enabled?

  1. High availability with a regional instance Correct answer.
  2. Automated backups retained for 30 days
  3. Read replicas in two other zones
  4. Binary logging with point-in-time recovery

Explanation

Configuring the instance as regional creates a standby in a second zone, and Cloud SQL promotes it automatically when the primary zone fails. Backups, binary logs, and read replicas support recovery or read scaling but not automatic write failover.

Question 31Design scalable and highly available cloud database solutions

An application issues many small single-region transactions against Spanner and needs the lowest possible commit latency. Which configuration most directly achieves this?

  1. A nameless multi-region read-only configuration
  2. A multi-region configuration with leader placement in another continent
  3. A multi-region instance configuration
  4. A regional instance configuration Correct answer.

Explanation

A regional Spanner configuration keeps all replicas and the leader in one region, so commits avoid cross-region quorum latency. Multi-region configurations add inter-region round trips to each commit, raising write latency.

Question 32Design scalable and highly available cloud database solutions

A Cloud SQL for PostgreSQL instance is used for both transactional writes and heavy reporting reads. Which combination best separates the workloads while keeping the primary protected?

  1. Add two read replicas and disable high availability
  2. Enable high availability only and run reports on the primary
  3. Enable high availability and add a read replica for reporting Correct answer.
  4. Use automated backups and run reports on the primary

Explanation

High availability provides an automatic standby for failover, while a read replica offloads reporting queries from the primary. Combining both separates read and write traffic without sacrificing zonal resilience.

Question 33Design scalable and highly available cloud database solutions

A team needs a globally distributed SQL database with external consistency for a financial ledger that spans continents. Which Google Cloud service meets these requirements?

  1. Spanner with a multi-region configuration Correct answer.
  2. Firestore in Datastore mode
  3. Bigtable with multi-cluster routing
  4. Cloud SQL for MySQL with cross-region replicas

Explanation

Spanner provides globally distributed SQL with external consistency, guaranteeing that transactions appear serialized in real time. Cloud SQL replicas are asynchronous, and the NoSQL options do not provide the same globally consistent SQL semantics.

Question 34Design scalable and highly available cloud database solutions

A Bigtable table uses monotonically increasing timestamp strings as row keys, and writes are concentrating on a single tablet. Which key design change distributes writes across tablets?

  1. Increase the number of nodes in the cluster
  2. Store the timestamp as a binary value
  3. Prefix each key with a hash of the timestamp Correct answer.
  4. Append a random suffix to each key

Explanation

Bigtable sorts rows lexicographically, so sequential keys land on the same tablet. Hashing a prefix spreads keys across the row space so writes distribute across tablets.

Question 35Design scalable and highly available cloud database solutions

An AlloyDB for PostgreSQL primary is saturated by read traffic from an analytics dashboard. Which approach adds read capacity without routing queries to the primary?

  1. Increase the primary's vCPU count
  2. Enable automated backups on the primary
  3. Move the dashboard queries into a stored procedure
  4. Create a read pool and direct dashboard queries to it Correct answer.

Explanation

AlloyDB read pools add independent read nodes that serve queries without burdening the primary. Scaling the primary or restructuring queries does not separate read traffic from the write node.

Question 36Design scalable and highly available cloud database solutions

A Spanner read-heavy workload is limited by CPU in the leader region while replicas in other regions are underused. Which change most directly adds read capacity without moving the leader?

  1. Enable automatic backups on the instance
  2. Switch to a single-region configuration
  3. Increase the leader region's node count
  4. Add read-only replicas in additional regions Correct answer.

Explanation

Read-only replicas serve strongly consistent reads locally and can absorb read traffic, reducing CPU pressure on the leader. Adding nodes to the leader increases write capacity but does not offload read processing to other regions.

Question 37Design scalable and highly available cloud database solutions

A Cloud SQL for MySQL read replica lags several minutes behind the primary, and reports require fresher data. Which action best restores acceptable replication freshness?

  1. Promote the replica to a standalone primary
  2. Take a backup of the replica
  3. Reduce the write rate on the primary or increase replica capacity Correct answer.
  4. Enable high availability on the replica

Explanation

Replica lag typically results from the primary applying writes faster than the replica can replay them. Lowering write pressure or giving the replica more resources lets it catch up.

Question 38Design scalable and highly available cloud database solutions

A Firestore application needs the strongest consistency for multi-document transactions across collections. Which configuration provides this?

  1. Firestore Native mode with multi-region location Correct answer.
  2. Firestore Native mode with a single-region location and eventual reads
  3. Firestore with a custom cache layer in front of reads
  4. Firestore in Datastore mode with ancestor queries only

Explanation

Firestore Native mode supports ACID transactions across documents, and a multi-region location keeps the database available across zones. Adding a cache layer or restricting to eventual reads weakens consistency.

Question 39Design scalable and highly available cloud database solutions

A team wants a managed PostgreSQL-compatible database that scales read traffic with additional nodes and offers a columnar engine for analytics. Which service should be chosen?

  1. AlloyDB for PostgreSQL Correct answer.
  2. Cloud SQL for PostgreSQL
  3. Spanner with PostgreSQL dialect
  4. Bigtable

Explanation

AlloyDB for PostgreSQL provides read pools for scaling reads and a columnar engine that accelerates analytical queries. Cloud SQL lacks the columnar engine, and the other options are not PostgreSQL-compatible in the same way.

Question 40Design scalable and highly available cloud database solutions

An AlloyDB read pool serves analytics queries while the primary handles writes. How does this differ from a Cloud SQL read replica?

  1. Cloud SQL read replicas share storage with the primary
  2. AlloyDB read pools can accept writes
  3. AlloyDB read pools use a shared storage layer so read nodes do not replay a write log Correct answer.
  4. Cloud SQL read replicas are always in the same zone as the primary

Explanation

AlloyDB separates compute from a distributed storage layer, so read pool nodes access the same storage without replaying a log, which reduces lag. Cloud SQL replicas maintain their own storage and apply the primary's binary log.

Question 41Design scalable and highly available cloud database solutions

A Cloud SQL for PostgreSQL instance must fail over automatically when its primary zone is lost, and the team also wants a cross-region copy for disaster recovery. Which combination meets both goals?

  1. A single-zone instance plus automated backups
  2. A single-zone instance plus a same-zone replica
  3. A regional instance only
  4. A regional instance plus a cross-region read replica Correct answer.

Explanation

A regional instance provides a standby in another zone for automatic failover, and a cross-region read replica provides a copy that can be promoted for regional disaster recovery. Neither feature alone covers both scenarios.

Question 42Design scalable and highly available cloud database solutions

A Spanner-backed application must stay available when its primary region becomes unreachable. Which practice most improves availability?

  1. Use a single-region configuration for lower latency
  2. Increase the node count in the primary region
  3. Use a multi-region configuration with replicas in separate regions Correct answer.
  4. Enable automatic backups on the instance

Explanation

A multi-region Spanner configuration keeps replicas in separate regions so the database can continue serving when one region is lost. Adding nodes or backups within one region does not protect against regional failure.

Question 43Design scalable and highly available cloud database solutions

An application writes to Bigtable with row keys that are sequential numeric IDs, and writes concentrate on one tablet. Which change best distributes the load?

  1. Reverse the byte order of the numeric key Correct answer.
  2. Switch the table to HDD storage
  3. Lower the replication factor of the cluster
  4. Increase the number of nodes in the cluster

Explanation

Sequential numeric keys sort together and create a hot tablet. Reversing the byte order spreads adjacent IDs across the row space, distributing writes across tablets.

Question 44Design scalable and highly available cloud database solutions

A workload requires very low-latency point lookups at petabyte scale with high write throughput and no relational joins. Which Google Cloud database is the best fit?

  1. Cloud SQL for PostgreSQL
  2. Bigtable Correct answer.
  3. Spanner
  4. Firestore in Native mode

Explanation

Bigtable is designed for petabyte-scale, low-latency key-based access with high write throughput and does not impose relational joins. The other options are relational or document databases with different scaling characteristics.

Question 45Design scalable and highly available cloud database solutions

A Cloud SQL for PostgreSQL instance must support high availability and read scaling for reporting. Which combination achieves both?

  1. A single-zone instance with automated backups
  2. A regional instance with no replicas
  3. A single-zone instance with a same-zone replica
  4. A regional instance with one or more read replicas Correct answer.

Explanation

A regional instance provides automatic failover through a standby, and read replicas offload reporting queries from the primary. The other combinations either lack failover or lack read scaling.

Question 46Design scalable and highly available cloud database solutions

A team is designing a Spanner schema for a multi-tenant SaaS application and wants to avoid hotspots from monotonically increasing tenant IDs. Which key strategy helps?

  1. Use a timestamp as the leading key column
  2. Use an auto-incrementing integer as the key
  3. Use the tenant ID directly as the first key column
  4. Hash the tenant ID and prefix it to the key Correct answer.

Explanation

Hashing the tenant ID before prefixing spreads keys across the key space, avoiding concentration on a single split. Direct monotonic values such as auto-incrementing integers or timestamps create hotspots.

Question 47Design scalable and highly available cloud database solutions

An AlloyDB cluster must accelerate analytical queries on transactional data without a separate warehouse. Which AlloyDB capability should be used?

  1. The columnar engine Correct answer.
  2. Connection pooling
  3. Read pools
  4. Automated backups

Explanation

The AlloyDB columnar engine stores data in an in-memory columnar format that speeds up analytical scans on transactional tables. Read pools add read nodes but do not provide columnar acceleration.

Question 48Design scalable and highly available cloud database solutions

A financial application uses Spanner and requires that a read never observes a transaction that committed later in real time. Which read type provides this guarantee without locking?

  1. A strong read at a timestamp in the past Correct answer.
  2. A read from a read-only replica without a timestamp
  3. A stale read at a timestamp in the past
  4. A read using a bounded staleness of 15 seconds

Explanation

A strong read returns the latest committed data and reflects all transactions committed before the read, providing external consistency. Stale and bounded-staleness reads intentionally return older data.

Question 49Design scalable and highly available cloud database solutions

A Cloud SQL for MySQL instance uses a regional configuration, and the team wants to test failover without disrupting production for long. Which approach is appropriate?

  1. Trigger a manual failover on the instance and observe the standby promotion Correct answer.
  2. Delete the standby and let Cloud SQL recreate it
  3. Disable high availability and re-enable it
  4. Restart the primary during peak traffic

Explanation

Cloud SQL supports a manual failover that promotes the standby, letting the team validate the failover path in a controlled way. Destructive or unplanned actions risk data loss or extended outages.

Question 50Design scalable and highly available cloud database solutions

A Bigtable workload requires reads to be served from the nearest cluster while writes remain consistent across clusters. Which feature should be enabled?

  1. Append-only writes with a random suffix
  2. A read-only cluster with no replication
  3. Multi-cluster routing with single-row transactions Correct answer.
  4. A single-cluster instance with SSD storage

Explanation

Bigtable multi-cluster routing directs reads to the nearest cluster while writes replicate across clusters with single-row transaction guarantees. A single cluster or a non-replicated cluster cannot serve reads from another location.