Microsoft announced the availability of Azure Managed Redis as a fully managed in-memory database service designed to accelerate applications and support modern cloud workloads and artificial intelligence applications, including retrieval-augmented generation (RAG) applications. The service combines the latest Redis innovations with provisioning, updates, patching, and automatic scaling, reducing the operational tasks required of development teams.
Azure Managed Redis stores data in memory rather than relying on disk-based storage, enabling data to be written and retrieved with lower latency. Microsoft says the service can provide high-performance access at a rate of up to millions of requests per second, with support for connections compatible with Redis clients.
Geographic Scaling and Performance Flexibility
The service supports availability of up to 99.999%, along with active geo-replication. Up to five Redis instances can be used to create a unified cache across Azure regions, with local primary replicas selected to optimize read and write performance globally.
Microsoft offers pricing options designed around memory and performance requirements, including workloads that require high compute density or large amounts of memory. The Flash optimized architecture also enables infrequently used data to be moved from RAM to NVMe storage, aiming to improve the handling of large datasets and reduce costs. Non-high-availability configurations are also available for development testing and performance evaluation.
Redis Capabilities for Data and AI Applications
Azure Managed Redis includes a range of Redis capabilities, including RediSearch, which turns the cache into a real-time search engine with querying across multiple fields, autocomplete, geospatial filtering, and k-nearest neighbor (KNN) search. These capabilities can be used in advanced artificial intelligence applications.
The service also supports RedisJSON for storing, querying, and updating data in JSON format; RedisTimeSeries for processing large volumes of time-series data; and RedisBloom for using probabilistic data structures. The use cases mentioned include vector similarity search, conversation-history storage, semantic caching for RAG applications, as well as session-data management and real-time leaderboards, and functioning as a publish/subscribe message broker.
Azure Managed Redis supports the following Redis data types:
- Strings for key-value pairs.
- Hashes for groups of key-value pairs.
- Lists, sets, and sorted sets.
- Bitmaps and HyperLogLog for approximating the number of unique items.
- Geospatial indexes and Streams for record-based data sequences.
Integration, Security, and Migration from the Previous Service
The service integrates with Azure App Service, Azure Functions, Azure SQL, Azure OpenAI Service, Azure Database for PostgreSQL, and Azure Kubernetes Service. Microsoft cites network isolation capabilities, authentication through Microsoft Entra ID, and built-in monitoring, in addition to the benefits of secure cloud infrastructure and its global network. The page also refers to more than 100 compliance certifications in Azure, including more than 50 certifications associated with specific regions and countries.
Microsoft explains that Azure Managed Redis is a first-party service developed in partnership with Redis, and that it is generally available and ready for production workloads worldwide. In parallel, the company announced the retirement of Azure Cache for Redis and recommended using Azure Managed Redis for existing and new workloads.
The product page contains a conflict regarding the retirement date for the Enterprise tier: in one place, it states that Azure Cache for Redis Enterprise will be retired on March 30, 2028, while elsewhere it gives the date as March 30, 2027. For the Basic, Standard, and Premium tiers, the page indicates that they will be retired on September 30, 2028. Microsoft says caches that are not migrated to Azure Managed Redis will be disabled after the retirement date, so the deadline for the Enterprise tier should be verified directly in Azure documentation before planning the migration.