Database pricing
AWS offers purpose-built managed databases across relational SQL, serverless NoSQL, search and analytics, document stores, graph engines, and in-memory caching. Google Cloud SQL and Memorystore sit beside them, grouped by the same database kinds.
Relational
Traditional managed relational database engines and distributed cloud-native relational databases.
AWS
Amazon RDS
Standard managed relational databases across six engines — PostgreSQL, MySQL, MariaDB, Oracle, SQL Server, and IBM Db2 — with dedicated EBS storage and Multi-AZ standby replication.
- 6Engines
- 37Regions
Amazon Aurora
High-performance MySQL- and PostgreSQL-compatible database with distributed 6-way storage replication, up to 15 read replicas, Serverless v2, and I/O-Optimized pricing.
- 2Engines
- 15Max Replicas
- 36Regions
Google Cloud
Key-value and NoSQL
Document, key-value and graph databases purpose-built for specialized access patterns and flexible schemas.
AWS
Also on AWS: DynamoDB, DocumentDB, Neptune →
Amazon DynamoDB
Fully managed, serverless key-value and document database with single-digit millisecond latency. On-demand and provisioned throughput, Standard vs Standard-IA storage, and Global Tables.
- 2Capacity Modes
- 2Table Classes
- 36Regions
Amazon DocumentDB
Fully managed, MongoDB-compatible JSON document database. Decoupled compute and auto-scaling 6-way replicated storage up to 128 TiB, with instance-based clusters and Elastic Clusters.
- MongoDB 5.0Compatibility
- 128 TiBStorage Scale
- 15Max Replicas
Amazon Neptune
Fully managed graph database engine optimized for relationship-dense queries. Native support for openCypher, Gremlin, and SPARQL, with Neptune Serverless and memory-optimized Neptune Analytics.
- 3Graph Models
- 1–128 NCUServerless Range
- 128 TiBMax Storage
Google Cloud
Not covered yet
No Google Cloud databases of this kind have been collected yet.
Cache
Sub-millisecond latency in-memory data stores and caching layers for read-heavy workloads.
AWS
Google Cloud
Analytics
Columnar data warehouses and managed search and log-analytics engines.
AWS
Also on AWS: Redshift, OpenSearch →
Amazon Redshift
Columnar SQL data warehouse. Provisioned RA3, RG (Graviton) and DC2 clusters billed per node-hour plus managed storage, or Serverless billed per RPU-hour while queries run.
- 10Node Types
- RPU-hrServerless
- 36Regions
Amazon OpenSearch Service
Managed OpenSearch and Elasticsearch-compatible search and log-analytics engine. Node-hour pricing for data, master, and UltraWarm tiers, plus EBS storage, Serverless OCUs, and Ingestion.
- 237Node Types
- 8Storage Options
- 36Regions
Google Cloud
Not covered yet
No Google Cloud databases of this kind have been collected yet.
AWS
Database architecture and pricing matrix
Also on AWS: database architecture and pricing matrix →
| Service | Primary Paradigm | Engine / API Compatibility | Compute & Scaling | Storage Architecture | Key Pricing Drivers |
|---|---|---|---|---|---|
| Amazon RDS | Relational (SQL) | PostgreSQL, MySQL, MariaDB, Oracle, SQL Server, IBM Db2 | Vertical instance resize, Read Replicas, Multi-AZ Standby | Dedicated EBS (gp3, io2, standard) up to 64 TiB | Instance $/hr + EBS $/GB-mo + IOPS/throughput tiers |
| Amazon Aurora | Distributed Relational (SQL) | PostgreSQL, MySQL | Serverless v2 (0.5–128 ACUs) or instances + up to 15 replicas | Decoupled 6-way replicated storage up to 128 TiB | ACU-hr or Instance $/hr + Storage $/GB-mo + I/O requests |
| Amazon Redshift | Data Warehouse (columnar SQL) | PostgreSQL-based SQL, Spectrum queries on S3 | Node count and type, Concurrency Scaling, or Serverless RPUs | Managed storage (RA3, RG, Serverless) or local SSD (DC2) | Node $/hr + Storage $/GB-mo, or RPU-hr + Storage $/GB-mo |
| Amazon DynamoDB | Key-Value & Document | DynamoDB API, PartiQL (SQL-compatible query) | Serverless on-demand or provisioned auto-scaling | SSD distributed serverless (Standard & Standard-IA) | On-Demand (WRU/RRU) or Provisioned (WCU/RCU) + Storage $/GB-mo |
| Amazon OpenSearch Service | Search & Analytics | OpenSearch API, Elasticsearch 7.x APIs, OpenSearch Dashboards | Instances + UltraWarm, or Serverless OCUs | EBS gp2/gp3/PIOPS or UltraWarm managed storage | Node $/hr + EBS/UltraWarm $/GB-mo, or OCU-hr |
| Amazon DocumentDB | Document Store (JSON/BSON) | MongoDB 3.6, 4.0, 5.0 API | Instances (db.r5/r6g/t4g) + 15 replicas, or Elastic Clusters | Decoupled 6-way replicated storage up to 128 TiB | Instance $/hr + Storage $/GB-mo + I/O (or I/O-Optimized flat) |
| Amazon Neptune | Graph Database | openCypher, Apache TinkerPop Gremlin, SPARQL 1.1 | Neptune Serverless (1–128 NCUs), instances + 15 replicas | Decoupled 6-way replicated storage up to 128 TiB | NCU-hr or Instance $/hr + Storage $/GB-mo + I/O requests |
| Amazon ElastiCache | In-Memory Key-Value / Cache | Valkey, Redis OSS, Memcached | Horizontal cluster sharding & vertical node resizing | Volatile / persistent in-memory RAM | Node $/hr across memory-optimized and burstable cache types |