Google Cloud · Managed service
Google AlloyDB for PostgreSQL
Relational
Google Cloud; PostgreSQL-compatible service
When to consider it
Consider AlloyDB for a PostgreSQL application whose measured transactional or mixed query workload justifies evaluating a specialized managed service. Compare it directly with Cloud SQL using the same queries.
What to check
PostgreSQL compatibility does not make every extension or operating procedure identical. Check supported features, instance sizing, migration effort, and total service cost.
Official documentation ↗Google Cloud · Analytics platform
Google BigQuery
Analytics
Google Cloud; managed analytical warehouse
When to consider it
Consider BigQuery for shared SQL analysis over large datasets, recurring reporting, and data engineering workloads. It separates the analytical job from the database serving application transactions.
What to check
Control scans, partition data around actual queries, and evaluate the chosen compute pricing model. Ingestion freshness, workload concurrency, and transfer costs belong in the design.
Official documentation ↗Google Cloud · Managed service
Google Cloud Bigtable
Wide-column / Time series
Google Cloud; distributed wide-column store
When to consider it
Evaluate Bigtable for large key-oriented datasets, high ingestion rates, and access over ordered row-key ranges. Telemetry and event histories can fit when their query patterns drive the key design.
What to check
Row-key design is central: hot ranges and unbounded scans can dominate behavior. Wide-column storage is different from an analytical columnar warehouse and does not provide a relational join model.
Official documentation ↗Google Cloud · Managed service
Google Cloud Firestore
Document
Google Cloud; document database with distinct editions and modes
When to consider it
Useful for applications organized around documents, including web and mobile backends. Its managed model and client-facing capabilities can reduce the amount of backend infrastructure you operate.
What to check
Choose the edition and mode deliberately. Query indexes, transaction limits, security rules, read charges, and client access patterns matter; do not treat a document store as an automatic substitute for relational joins.
Official documentation ↗Google Cloud · Managed service
Google Cloud Memorystore
Cache / Key-value
Google Cloud; managed Valkey, Redis, and Memcached offerings
When to consider it
A managed option for caching hot records, session access, and other memory-oriented operations alongside an authoritative database. Select the particular engine and deployment that match the application.
What to check
Command support, persistence, failover, and clustering vary by offering. Define eviction and cache rebuilding; a fast managed cache does not remove invalidation or durability decisions.
Official documentation ↗Google Cloud · Managed service
Google Cloud Spanner
Relational / Distributed SQL
Google Cloud; distributed database with GoogleSQL and PostgreSQL dialects
When to consider it
A candidate for relational transactions that need horizontal distribution and coordinated consistency across a regional or multi-region deployment. Evaluate it when a single primary database becomes a demonstrated constraint.
What to check
Distributed transactions add coordination. Model keys and locality carefully, and verify SQL dialect, feature support, regional configuration, and cost before assuming an existing PostgreSQL application will transfer unchanged.
Official documentation ↗Google Cloud · Managed service
Google Cloud SQL
Relational
Google Cloud; PostgreSQL, MySQL, or SQL Server engine
When to consider it
A managed home for applications using established relational engines. Consider it when you want conventional SQL, transactions, and familiar drivers while Google handles infrastructure maintenance.
What to check
Select the engine first. Extensions, administrative access, high availability, supported versions, and recovery options differ from an unrestricted self-managed installation.
Official documentation ↗