AWS MemoryDB

Durable in-memory database service compatible with Valkey and Redis OSS, delivering microsecond read latency and single-digit millisecond writes. Unlike traditional in-memory caches, it uses a Multi-AZ distributed transaction log to persist data across availability zones, making it suitable as a primary database for workloads that need both speed and durability.

AWS MemoryDB supports millions of requests per second at scale, with built-in high availability targeting 99.99% uptime and near-instant recovery without data loss. It distinguishes itself from cache-only offerings by combining the familiar Valkey and Redis OSS API with enterprise-grade durability, so teams can use existing drivers, tooling, and application code without changes.

Key capabilities:

• Valkey and Redis OSS API compatibility with streams, sets, lists, and common data structures
• Multi-AZ durability via distributed transactional log, eliminating the need for separate cache and database layers
• Vector search for retrieval-augmented generation, recommendations, and semantic search in generative AI workflows
• Native JSON support with JSONPath queries and partial document updates
• Kubernetes integration via AWS Controllers for Kubernetes for defining and managing clusters from within clusters

Common workloads include real-time gaming session stores and leaderboards, high-concurrency streaming and media feeds, web and mobile application backends requiring low latency, and generative AI applications that need fast vector search with high recall. Developers typically connect using standard Valkey or Redis OSS clients with minimal code changes.