Compare database services across AWS, Azure, and GCP including relational databases, NoSQL databases, pricing, performance, and use cases.
Recommendation: AWS is recommended for enterprises needing the broadest selection of purpose-built database engines and mature tooling. Azure excels for SQL Server migrations with Hybrid Benefit and globally distributed multi-model NoSQL via Cosmos DB. GCP is ideal for globally consistent relational databases (Cloud Spanner), real-time mobile/web applications (Firestore), and high-performance PostgreSQL workloads (AlloyDB).
| Category | AWS | Azure | GCP |
|---|---|---|---|
| Relational Database Engines Supported relational database engines and variants | Aurora (MySQL/PostgreSQL), MySQL, PostgreSQL, MariaDB, Oracle, SQL Server | SQL Database, Database for MySQL, Database for PostgreSQL, SQL Managed Instance | Cloud SQL (MySQL, PostgreSQL, SQL Server), AlloyDB, Cloud Spanner |
| NoSQL Data Models Supported NoSQL data models and APIs | Key-value, document (DynamoDB), graph (Neptune), wide-column | Document, key-value, graph, column-family, table (Cosmos DB multi-model) | Document (Firestore), wide-column (Bigtable), key-value |
| Global Distribution Multi-region and global database distribution capabilities | DynamoDB Global Tables, Aurora Global Database (up to 5 secondary regions) | Cosmos DB multi-region writes with turnkey global distribution in 60+ regions | Cloud Spanner provides global relational distribution with 99.999% SLA |
| Automatic Scaling Ability to automatically scale storage and compute | Aurora auto-scales storage to 128 TB; DynamoDB on-demand scales automatically | SQL Database Hyperscale auto-scales; Cosmos DB autoscale throughput | Firestore scales automatically; Cloud Spanner scales compute independently |
| Serverless Option Serverless database offerings that scale to zero | Aurora Serverless v2, DynamoDB on-demand mode | Azure SQL Database Serverless, Cosmos DB serverless | Firestore (fully serverless), Cloud SQL has no serverless option |
| Max Database Size Maximum supported database size for managed relational databases | 128 TB (Aurora), 64 TB (RDS) | 100 TB (SQL Hyperscale) | Unlimited (Cloud Spanner), 64 TB (Cloud SQL) |
| In-Memory Caching Managed in-memory caching services | ElastiCache (Redis, Memcached), DynamoDB DAX | Azure Cache for Redis | Memorystore (Redis, Memcached) |
| Graph Database Managed graph database for relationship-heavy data | Amazon Neptune | Cosmos DB Gremlin API | ✗ No |
| Consistency Models Available consistency levels for NoSQL databases | DynamoDB: eventual and strong consistency | Cosmos DB: strong, bounded staleness, session, consistent prefix, eventual | Firestore: strong consistency for all reads; Spanner: external consistency |
| Real-Time Sync Built-in real-time data synchronization for client applications | ✗ No | Cosmos DB Change Feed | Firestore real-time listeners with offline support |
Pricing: RDS starts at $0.017/hr (db.t4g.micro). Aurora starts at $0.073/hr (db.r6g.large). DynamoDB on-demand: $1.25 per million write requests, $0.25 per million read requests. Free tier includes 750 hours RDS and 25 GB DynamoDB.
Performance: Aurora delivers up to 5x MySQL throughput. DynamoDB provides consistent single-digit millisecond latency. ElastiCache delivers sub-millisecond response times for caching workloads.
Pricing: SQL Database starts at $4.90/month (Basic tier). Cosmos DB: $0.25 per million RUs (serverless). Elastic pools start at $0.0023/eDTU/hr. Azure Hybrid Benefit applies to SQL workloads.
Performance: SQL Hyperscale delivers rapid scale-out with up to 100 TB. Cosmos DB guarantees <10ms reads and <15ms writes at 99th percentile globally. Intelligent Query Processing optimizes SQL workloads automatically.
Pricing: Cloud SQL starts at $0.0150/hr (db-f1-micro). Firestore: $0.06 per 100K reads, $0.18 per 100K writes. Cloud Spanner: $0.90/node/hr. AlloyDB starts at $0.1388/hr.
Performance: AlloyDB delivers up to 4x faster transactional and 100x faster analytical queries than standard PostgreSQL. Cloud Spanner provides globally consistent reads with single-digit millisecond latency. Firestore scales to millions of concurrent clients.
AWS offers the broadest selection of purpose-built databases with Amazon RDS for relational workloads and DynamoDB for NoSQL at any scale.
Amazon RDS supports six database engines: Aurora (MySQL/PostgreSQL compatible), MySQL, PostgreSQL, MariaDB, Oracle, and SQL Server. Aurora delivers up to 5x throughput of standard MySQL and 3x of PostgreSQL with automatic storage scaling up to 128 TB.
DynamoDB is a fully managed NoSQL database delivering single-digit millisecond performance at any scale. Supports key-value and document data models with built-in security, backup, restore, and in-memory caching via DAX.
AWS provides purpose-built databases including ElastiCache (in-memory), Neptune (graph), Timestream (time-series), QLDB (ledger), and DocumentDB (MongoDB compatible) for specialized workloads.