Amazon MemoryDB
aws.amazon.com
- Web
- API
What Amazon MemoryDB is
Amazon MemoryDB is a paid cloud database service compatible with Valkey and Redis OSS. It is designed as a durable primary database for real-time generative AI, web and mobile applications, online games, and media streaming. MemoryDB keeps the full dataset in memory; AWS describes microsecond reads and single-digit millisecond writes. A distributed transactional log and storage across multiple Availability Zones support durability, consistency, and recovery. Valkey and Redis OSS data types, parameters, and commands are supported, except administrative commands. Clusters scale through node changes, sharding for write capacity, and replicas for read capacity. Multi-Region provides active-active replication and up to 99.999% availability, but does not support data tiering or vector search. Users can manage clusters through the console, CLI, or SDK, with CloudWatch metrics and automatic patching. The service runs in a VPC, supports KMS encryption at rest and TLS in transit, and uses paid node-hour pricing with additional data-written charges.
Who it suits
It suits workloads needing a durable, very fast Valkey- or Redis OSS-compatible primary database. It is intended for cloud applications that can use the listed protocols and clustering options.
Good to know, and what to watch
- Valkey and Redis OSS compatibility
- Microsecond read latency stated
- Multi-Availability Zone storage
- Sharding and replicas for scaling
- KMS and TLS encryption
- No free plan
- Data tiering has workload limits
- Multi-Region lacks vector search
- Additional charges apply for data written
Plans
Every plan and priceOn-demand nodes
Reserved nodes
The verdict
MemoryDB targets durable, in-memory primary databases with Valkey and Redis OSS compatibility. Review protocol fit, Multi-Region limitations, and node, data-written, snapshot, and transfer charges before adoption.
What the maker publishes
- AWS company
- Amazon Web Services says it launched in 2006 and is part of Amazon. source
- Compatibility
- MemoryDB supports Valkey and Redis OSS data types, parameters, and commands, with the FAQ noting that admin commands are excluded. source
- Data tiering limit
- AWS says data tiering is intended for workloads that regularly access up to 20% of their data and can tolerate extra latency on first access to less frequently used items. source
- Durability
- MemoryDB uses a distributed transactional log and stores data across multiple Availability Zones for durability, consistency, and recovery. source
- Management and integrations
- Users can create clusters with the AWS Management Console, CLI, or SDK, and MemoryDB provides Amazon CloudWatch metrics and automatic software patching. source
- Multi-Region
- MemoryDB Multi-Region supports active-active replication and up to 99.999% availability; AWS says it does not support data tiering or vector search. source
- Performance
- AWS says MemoryDB stores the full dataset in memory and provides microsecond read latency and single-digit millisecond write latency. source
- Pricing
- On-demand nodes are billed by instance hour; data written is priced at $0.20/GB, while reads have no associated data-written charge. source
- Scaling
- Clusters can scale by changing node counts or node types, and by using sharding for write scaling and replicas for read scaling. source
- Security
- MemoryDB runs in Amazon VPC, supports encryption at rest through AWS KMS and encryption in transit with TLS, and Graviton2 nodes include always-on 256-bit DRAM encryption. source
- Use cases
- AWS describes use cases including real-time generative AI applications, web and mobile apps, online games, and media streaming. source
- What it does
- Amazon MemoryDB is a Valkey- and Redis OSS-compatible durable in-memory database service. source
- Who it is for
- AWS positions MemoryDB for workloads needing a durable, ultra-fast Valkey- or Redis OSS-compatible primary database. source
Checked against the maker's pages on 8 Oct 2026.