Databases / Databases

Explore the database landscape.

Compare 49 database engines, managed services, and analytical platforms across Google Cloud, Microsoft and Azure, Amazon Web Services, Oracle, and the wider ecosystem.

Start with a provider, or compare across them.

An engine is the database software. SQL Server, PostgreSQL, Oracle Database, and MySQL can run in several environments.

A managed service is an operating choice. RDS and Cloud SQL host familiar engines; Aurora and AlloyDB have their own compatibility and architecture boundaries.

A specialist has a specific job. BigQuery, Redshift, and Fabric Warehouse serve analytics. Search indexes and caches need their own data ownership and recovery plan.

These are distinct product entries, not a count of unrelated database engines or a popularity ranking. Coverage is curated, not exhaustive. Product sources reviewed September 8, 2026. How we evaluate products →

Clear filters

49 of 49 engines and services · Alphabetical order

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 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 ↗

Continue your work

Take the next step.