Amazon Web Services · Managed service
Amazon Aurora
Relational
AWS; MySQL-compatible or PostgreSQL-compatible database
When to consider it
Evaluate Aurora for relational applications on AWS that fit one of its supported compatibility models. Compare it with RDS for the same engine, using realistic read/write patterns and recovery requirements.
What to check
Aurora is not identical to stock MySQL or PostgreSQL. Check engine features, extension support, I/O pricing, replica behavior, and regional write topology.
Official documentation ↗Amazon Web Services · Managed service
Amazon Aurora DSQL
Relational / Distributed SQL
AWS; serverless distributed SQL with PostgreSQL compatibility
When to consider it
A candidate for distributed transactions with concurrent regional read/write endpoints and strong consistency. Consider it when the selected AWS region set and supported SQL features match a new or adaptable application.
What to check
DSQL is distinct from Aurora PostgreSQL. Validate supported SQL, transaction retries, snapshot isolation, and region sets; current multi-region clusters do not span continents.
Official documentation ↗Amazon Web Services · Managed service
Amazon DocumentDB
Document
AWS; document database with MongoDB compatibility
When to consider it
Evaluate DocumentDB when a document workload belongs in AWS and its supported MongoDB-compatible interfaces cover the application. Include a comparison with MongoDB Atlas and a relational JSON approach.
What to check
MongoDB compatibility does not mean it runs MongoDB or supports every feature. Test operators, indexes, transactions, drivers, and migration tools against the exact DocumentDB engine version.
Official documentation ↗Amazon Web Services · Managed service
Amazon DynamoDB
Key-value / Document
AWS; managed key-value and document database
When to consider it
Useful for predictable key-oriented access patterns where partition design and managed scaling suit the application. Design primary keys and secondary indexes from the reads the application must perform.
What to check
Hot keys, indexes, and consistency choices affect the design. Global table consistency modes and index read guarantees differ; check the exact API and configuration.
Official documentation ↗Amazon Web Services · Managed service
Amazon ElastiCache
Cache / Key-value
AWS; managed Valkey, Redis OSS, and Memcached offerings
When to consider it
Useful for offloading repeated reads, session access, and other low-latency cache operations from a primary database. Select an engine and deployment based on actual client and data requirements.
What to check
Treat eviction, persistence, failover, and cache rebuilding as explicit choices. ElastiCache and durable database services serve different reliability needs.
Official documentation ↗Amazon Web Services · Managed service
Amazon Neptune
Graph
AWS; managed graph database
When to consider it
Consider Neptune when graph traversals or RDF queries are central and the selected graph model and query language match the application. Fraud relationships and connected-data queries are common evaluation scenarios.
What to check
Property graphs and RDF are different modeling choices. Test supported query languages, highly connected nodes, ingestion, and the regional deployment topology.
Official documentation ↗Amazon Web Services · Managed service
Amazon OpenSearch Service
Search
AWS; managed search and analytics service
When to consider it
A candidate for indexed search, log exploration, and retrieval workloads in AWS. It provides a managed operating path for workloads that match the supported OpenSearch features.
What to check
Choose provisioned or serverless capabilities deliberately. Index mappings, shard design, ingestion, access controls, and reindexing still require ownership.
Official documentation ↗Amazon Web Services · Managed service
Amazon RDS
Relational
AWS; PostgreSQL, MySQL, MariaDB, SQL Server, Oracle Database, or Db2
When to consider it
A managed hosting choice when you want an established relational engine on AWS. It is especially relevant when engine compatibility and reducing routine infrastructure maintenance matter more than changing the data model.
What to check
RDS is a service, not a single database engine. Versions, licensing, extensions, administrative access, and availability options depend on the selected engine and configuration.
Official documentation ↗Amazon Web Services · Analytics platform
Amazon Redshift
Analytics
AWS; managed data warehouse with provisioned and serverless options
When to consider it
A warehouse candidate for SQL reporting, analytical joins, and data pipelines in AWS. Evaluate it when the workload needs shared analysis rather than a primary store for application transactions.
What to check
Compare query concurrency, data layout, ingestion, and compute choices. PostgreSQL-related SQL familiarity does not make Redshift a replacement for an OLTP PostgreSQL server.
Official documentation ↗Other platforms · Database engine
Apache Cassandra
Wide-column
Distributed engine; self-managed or hosted offerings
When to consider it
A candidate for large partitioned datasets with predictable key-based access and distributed writes. Query-driven modeling is essential to making its wide-column approach useful.
What to check
Do not expect relational joins, foreign keys, or general cross-partition transactions. Plan partition sizes, consistency settings, repairs, compaction, and node operations.
Official documentation ↗Microsoft & Azure · Managed service
Azure AI Search
Search
Microsoft Azure; search and retrieval service
When to consider it
Consider it for full-text, vector, and hybrid retrieval over content indexed from an authoritative source. It is useful when search relevance and Azure service integration are central to the application.
What to check
Plan document ingestion, index updates, capacity, and reindexing. Search results can lag source transactions; the search index should not silently become the system of record.
Official documentation ↗Microsoft & Azure · Managed service
Azure Cosmos DB
Document / Key-value
Microsoft Azure; distributed NoSQL service
When to consider it
A candidate for partitioned document or key-oriented application data when Azure integration and regional distribution matter. Choose the API, partition key, and consistency model around specific access patterns.
What to check
Request units and hot partitions affect cost and performance. Accounts with multiple write regions cannot use strong consistency; transactions and query behavior have API-specific boundaries.
Official documentation ↗Microsoft & Azure · Managed service
Azure Database for MySQL
Relational
Microsoft Azure; managed MySQL
When to consider it
A managed hosting option for MySQL applications in Azure. Consider it when retaining the MySQL engine and existing client libraries matters more than adopting SQL Server-specific features.
What to check
Check supported MySQL versions, migration restrictions, networking, high availability, and parameter access. Managed infrastructure does not remove query tuning or the need to rehearse recovery.
Official documentation ↗Microsoft & Azure · Managed service
Azure Database for PostgreSQL
Relational
Microsoft Azure; managed PostgreSQL
When to consider it
A practical option for PostgreSQL applications hosted in Azure that need managed patching, backup facilities, and integration with Azure networking and identity. Compare it with SQL Server-based services when engine compatibility is still undecided.
What to check
Check supported PostgreSQL versions, extensions, connection limits, and high-availability configuration. Managed hosting still leaves schema design, query tuning, and recovery validation with your team.
Official documentation ↗Microsoft & Azure · Managed service
Azure Managed Redis
Cache / Key-value
Microsoft Azure; managed Redis service
When to consider it
An Azure option for frequently accessed data, session storage, and cache-aside designs. It can reduce infrastructure work when an application already uses Redis-compatible client access.
What to check
Validate supported commands, tier capabilities, clustering, memory policies, and persistence. Keep critical records in an authoritative store unless the selected durability model has been explicitly justified.
Official documentation ↗Microsoft & Azure · Managed service
Azure SQL Database
Relational
Microsoft Azure; managed database service using SQL Server technology
When to consider it
A candidate for new or modernized applications that need managed relational storage in Azure. It offers a database-focused operating model while Microsoft manages the underlying platform.
What to check
It is not a full SQL Server instance under your control. Validate server-level features, cross-database dependencies, service-tier limits, and migration compatibility.
Official documentation ↗Microsoft & Azure · Managed service
Azure SQL Managed Instance
Relational
Microsoft Azure; managed SQL Server-compatible instance
When to consider it
Consider Managed Instance when an existing SQL Server estate needs broader instance-level compatibility than Azure SQL Database provides. It can reduce refactoring for applications with established database dependencies.
What to check
Near-full compatibility still requires an assessment. Review unsupported features, networking, provisioning, maintenance behavior, and the cost of the selected service configuration.
Official documentation ↗Other platforms · Database engine
ClickHouse
Analytics
Self-managed columnar engine or ClickHouse Cloud
When to consider it
Useful for large event aggregations and analytical serving workloads with sustained ingestion. Its columnar approach can fit queries that scan many records but only some fields.
What to check
Ordering, partitioning, ingestion batches, updates, and retention affect behavior. It should not casually replace a database designed around transactional application updates.
Official documentation ↗Other platforms · Database engine
CockroachDB
Relational / Distributed SQL
Distributed SQL engine; self-managed and managed options
When to consider it
Consider it when a relational workload needs distributed transactions and regional placement beyond one conventional primary. Test representative contention and locality along with the query model.
What to check
SQL compatibility, transaction retries, coordination latency, survival settings, and licensing need evaluation. Multiple regions are not automatically the right answer for global readers.
Official documentation ↗Other platforms · Database engine
Couchbase
Document / Key-value
Couchbase Server or managed Capella service
When to consider it
Consider Couchbase for document and key-value access with SQL-like document queries, especially when the wider mobile synchronization ecosystem is relevant. Validate the exact components needed.
What to check
Query indexes, service sizing, consistency, and synchronization are architectural choices. SQL-like syntax does not make document modeling equivalent to a relational schema.
Official documentation ↗Other platforms · Embedded engine
DuckDB
Analytics
Embedded analytical engine on a local machine or within an application
When to consider it
Useful for SQL analysis over files and local datasets without operating a warehouse service. It suits exploratory work and analytical processing that fits the surrounding machine.
What to check
Account for memory, storage, and concurrency. An embedded analytical engine is not automatically a shared, always-on application database.
Official documentation ↗Other platforms · Database engine
Elasticsearch
Search
Distributed search engine; self-managed and Elastic Cloud options
When to consider it
Useful for relevance-ranked retrieval, filtering, and aggregation over indexed data. Evaluate it when search and analysis requirements justify a dedicated indexing system.
What to check
Mappings, analyzers, shards, ingestion, and reindexing need ownership. Results may lag the source; identify the authoritative record store.
Official documentation ↗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 ↗Other platforms · Database engine
IBM Db2
Relational
IBM database family; platform-specific and managed offerings
When to consider it
An important option for existing IBM estates and enterprise applications certified for Db2. Identify the actual Db2 product and platform before comparing migration or operating options.
What to check
Db2 Database, Db2 for z/OS, and other family members are not one interchangeable deployment. Review platform support, licensing, SQL behavior, tooling, and application dependencies.
Official documentation ↗Other platforms · Database engine
InfluxDB
Time series
Time-series database; distinct versions and deployment products
When to consider it
Consider InfluxDB when metrics, telemetry, and time-oriented ingestion dominate. Match the chosen version and product to the query language, retention, and operating model you need.
What to check
InfluxDB generations and products differ. Verify query capabilities, clustering, retention, migration, and supported deployment before applying advice from another version.
Official documentation ↗Other platforms · Database engine
MariaDB
Relational
Self-managed server; managed offerings also available
When to consider it
An established relational engine worth considering where MariaDB tooling, application support, or existing operating knowledge fit the project. Evaluate it as its own product.
What to check
MySQL heritage does not guarantee identical features or upgrade behavior. Check SQL compatibility, storage engines, replication, extensions, and the exact release path.
Official documentation ↗Microsoft & Azure · Analytics platform
Microsoft Fabric Warehouse
Analytics
Microsoft Fabric; analytical warehouse integrated with OneLake
When to consider it
A candidate for teams building a Microsoft Fabric analytics environment with SQL warehousing and shared data workflows. Assess it alongside other warehouses using representative reporting and ingestion workloads.
What to check
Fabric Warehouse is distinct from SQL database in Fabric and from SQL Server OLTP. Evaluate T-SQL support, capacity sharing, ingestion design, and governance requirements.
Official documentation ↗Microsoft & Azure · Database engine
Microsoft SQL Server
Relational
Self-managed servers or cloud VMs; managed services also available
When to consider it
A core choice for existing T-SQL applications, Microsoft tooling, and enterprise workloads that depend on SQL Server features. Evaluate the installed engine separately from Azure services built on its technology.
What to check
Check edition, licensing, operating platform, and feature requirements. Stored procedures, server-level integrations, and administrative jobs can make migration more involved than moving tables.
Official documentation ↗Other platforms · Database engine
MongoDB
Document
Self-managed database or MongoDB Atlas managed service
When to consider it
A candidate when documents align with the records commonly read and changed together. MongoDB Atlas supplies a managed deployment option; that hosting choice is separate from the document model.
What to check
Define schema validation and document boundaries. Cross-document workflows, transactions, sharding, and secondary indexes still require deliberate design.
Official documentation ↗Oracle · Database engine
MySQL
Relational
Self-managed engine; many providers offer managed hosting
When to consider it
An established relational option for web applications and services with a MySQL-compatible ecosystem. Existing team experience, application support, and operational procedures may make it a practical first candidate.
What to check
Check the storage engine, SQL modes, transaction isolation, indexing, and migration requirements. MySQL, MariaDB, Aurora MySQL, and HeatWave should not be treated as interchangeable products.
Official documentation ↗Oracle · Managed service
MySQL HeatWave
Relational / Analytics
Managed MySQL with optional HeatWave capabilities; cloud offerings vary
When to consider it
Evaluate HeatWave when MySQL transactional data and accelerated analytical queries are useful together. Compare the selected service configuration with ordinary managed MySQL and a separate analytical warehouse.
What to check
A DB system and a HeatWave cluster are distinct parts of the configuration. Check the actual cloud offering, supported features, sizing, data loading, and cost.
Official documentation ↗Other platforms · Database engine
Neo4j
Graph
Property graph engine; self-managed or Neo4j Aura
When to consider it
A candidate when multi-hop relationship traversal is a central query rather than an occasional report. Evaluate the property graph model and Cypher against realistic connected data.
What to check
Highly connected nodes and unrestricted traversals can change query costs. Compare bounded relational joins before adding graph synchronization and another operating surface.
Official documentation ↗Oracle · Database engine
Oracle AI Database
Relational
Oracle Database engine; on-premises and cloud deployment options
When to consider it
A central candidate for existing Oracle applications, PL/SQL code, and enterprise systems that depend on Oracle database features. The current product name is Oracle AI Database; compatibility still depends on version and deployment.
What to check
Review edition and option licensing, application certification, operational skills, and migration dependencies. A new engine can require rewriting database-specific code and administration workflows.
Official documentation ↗Oracle · Managed service
Oracle Autonomous AI Database
Relational / Analytics
Oracle managed database service; deployment and workload choices vary
When to consider it
Consider it for Oracle SQL and PL/SQL workloads where managed patching, tuning, and database operations fit the application. Select a transactional or analytical workload configuration deliberately.
What to check
Automation does not establish application compatibility or recovery readiness. Check database restrictions, integration dependencies, licensing model, workload configuration, and export requirements.
Official documentation ↗Oracle · Managed service
Oracle NoSQL Database Cloud Service
Key-value / Document
Oracle Cloud; managed NoSQL tables and key-oriented access
When to consider it
An option for applications that fit key-oriented tables and flexible records in Oracle Cloud. Evaluate table design, throughput needs, and regional access patterns against the service model.
What to check
Do not assume relational joins or Oracle Database transaction semantics. Validate consistency, request limits, capacity choices, and global-table conflict behavior.
Official documentation ↗Other platforms · Database engine
PostgreSQL
Relational
Self-managed engine; many managed hosting services
When to consider it
A strong general-purpose candidate for related records, constraints, joins, and transactions. It can also support flexible JSON attributes without automatically introducing a separate document database.
What to check
Choose isolation, indexes, extensions, and replication deliberately. A familiar engine still needs query tuning, a tested recovery process, and a deployment that fits its write requirements.
Official documentation ↗Other platforms · Database engine
Redis
Cache / Key-value
Memory-oriented engine; self-managed and managed offerings
When to consider it
Useful for fast access to data structures, frequently reused records, and cache patterns. It can support more than caching, but the role and durability expectations must be explicit.
What to check
Eviction, persistence settings, replication, failover, and version-specific licensing matter. Do not infer durability from memory speed or a replica count.
Official documentation ↗Other platforms · Database engine
SAP HANA
Relational / Analytics
In-memory relational platform; on-premises and cloud offerings
When to consider it
Relevant to SAP application landscapes and workloads combining transactions and analysis on an in-memory relational system. Application certification and the selected deployment should drive the evaluation.
What to check
Memory sizing, persistence, recovery, licensing, and SAP integration requirements are material. HANA and HANA Cloud have deployment-specific capabilities and responsibilities.
Official documentation ↗Other platforms · Analytics platform
Snowflake
Analytics
Managed data platform on supported public clouds
When to consider it
A warehouse and data-platform candidate for shared analytics, data engineering, and workloads that benefit from independently managed compute resources. Compare it with cloud-native warehouses using your actual query mix.
What to check
Cloud, region, edition, warehouse sizing, and data movement influence cost and capabilities. Keep transactional application requirements separate from the analytics evaluation.
Official documentation ↗Other platforms · Embedded engine
SQLite
Relational
Embedded library; local database file
When to consider it
Useful for local applications, device storage, and workloads that can share one host and serialize writes. It avoids running a separate database server.
What to check
Do not directly share its file across networked application servers. Evaluate writer concurrency, file ownership, backup consistency, and the limits of the surrounding host.
Official documentation ↗Other platforms · Database extension
TimescaleDB
Time series / Relational
PostgreSQL extension; self-managed or compatible managed hosting
When to consider it
A candidate for timestamped records that need time-window queries, retention management, and joins with relational data. It extends PostgreSQL rather than introducing a wholly separate query engine.
What to check
Validate extension and hosting compatibility, partition intervals, retention policies, and ingestion behavior. Plain PostgreSQL may suffice for a smaller workload.
Official documentation ↗Other platforms · Database engine
Typesense
Search
Search engine; self-managed or Typesense Cloud
When to consider it
A candidate for application search that needs ranked text retrieval, typo tolerance, and filters over structured collections. Compare the relevant features with your existing database search.
What to check
Check indexing needs, data size, filtering, and update behavior. Keep a tested path to rebuild the search collection from its authoritative source.
Official documentation ↗Other platforms · Database engine
Valkey
Cache / Key-value
Open-source memory-oriented engine; self-managed or hosted
When to consider it
An option for memory-oriented key access and cache workloads when its data structures and client compatibility fit the application. Several clouds provide managed Valkey offerings.
What to check
Check commands, modules, persistence, and cluster behavior for the selected version. Compatibility with a Redis client does not imply every Redis feature is available.
Official documentation ↗