Build new caches on the fastest, most feature-rich Redis yet, fully managed and supported by Microsoft.
Author: Purna Mehta, Product Manager, Redis Inc and Shruti Pathak, Product Manager, Microsoft
Delivering responsive applications at scale takes more than a fast cache. Applications need to serve frequently accessed data without repeatedly querying backend databases, manage session and application data efficiently, and power relevant search and AI experiences. Redis brings these capabilities together in a low-latency data layer—helping teams improve responsiveness and simplify application development.
Today we're announcing the public preview of Redis 8 on Azure Managed Redis. If you're creating a new Azure Managed Redis instance, you can now develop against Redis 8 from day one: faster commands, more data structures built in, and new tools for streams, vectors and observability, all on Azure Managed Redis, the Microsoft-supported service built on Redis Enterprise.
Redis 8 is a big leap in performance in the history of Redis, and it ships as "one Redis": the capabilities that used to require separate modules are now part of the core product. Whether you're building a cache, a session store, a real-time leaderboard or an AI app with semantic search, you now start with the complete toolkit.
What does Redis 8 mean for your applications and enterprises
For developers, IT decision-makers, and business leaders, the question is how these capabilities support application and operational priorities. The following sections explain what to evaluate, from application responsiveness and development effort to the complexity of managing application data.
Performance improvements in Redis8. Redis 8 includes more than 30 performance improvements. In Redis's own benchmarks against Redis 7.2.5, p50 command latency dropped by up to 87%, and the new I/O threading implementation delivered up to 2x more operations per second. Note: These are benchmark results published by Redis Inc.; your results on Azure Managed Redis will vary by workload and tier.
Bring all Redis capabilities together. Time series, probabilistic structures (Bloom filter, cuckoo filter, count-min sketch, top-k, t-digest), JSON capabilities, the Redis Query Engine (formerly RediSearch) and vector set ship together. You can start using a module as the need arises, instead of having to configure modules at ahead of time.
Simplify session management and application logic. Hash field expiration, introduced in Redis 7.4, gets three new commands in Redis 8: HGETEX, HSETEX and HGETDEL. They replace multi-step read-then-expire patterns with a single atomic call, which is exactly what session and cache-aside code needs.
Support richer search and AI experiences.. Vector set and the Redis Query Engine(formerly RediSearch) let you run semantic search, recommendations and RAG retrieval in the same database that already serves your cache. Fewer systems to deploy, secure and pay for.
Your existing tools keep working. The official client libraries (redis-py, node-redis, Jedis, Lettuce, go-redis, NRedisStack) and Redis Insight support Redis 8.
Match Redis 8 capabilities to your application priorities
|
Capability |
What it gives you |
Try it for |
|
Hash field expiration + HGETEX, HSETEX, HGETDEL |
Per-field TTLs and atomic get-and-expire |
Session stores, cache-aside, rate limits |
|
JSON |
Store documents and update paths atomically with JSONPath |
User profiles, carts, feature flags |
|
Redis Query Engine |
Secondary indexes, full-text and vector search, aggregations |
Product search, RAG retrieval |
|
Vector set (beta) |
Native vector similarity data type |
Recommendations, semantic caching |
|
Streams: XDELEX, XACKDEL (8.2) |
Acknowledge and delete entries atomically, with consumer-group control |
Event pipelines, work queues |
|
Probabilistic structures |
Bloom, cuckoo, count-min sketch, top-k, t-digest in small memory |
Deduplication, "seen before" checks, trending lists |
|
Time series |
Fast ingest, compaction, range queries |
Metrics, IoT telemetry |
|
CLUSTER SLOT-STATS (8.2) |
Per-slot key count, CPU and network usage |
Finding hot keys and uneven shards |
|
Keyspace events OVERWRITTEN, TYPE_CHANGED (8.2) |
Finer-grained change notifications |
Reactive apps, cache invalidation |
Redis 8.0 changed ACL behavior so module commands fall under standard categories like @read and write; review your ACL rules if you're porting an app. See the Redis 8.0 and Redis 8.2 release notes for details.
What enterprise teams should evaluate in the preview Redis 8.2 on Azure Managed Redis is in public preview and is available for new instances only. You can't yet upgrade an existing Redis 7.4 cache to Redis 8.2 in place, so for now it's best suited to new development, proofs of concept and testing. Existing Redis 7.4 caches keep running as before. Redis 8.2 is currently not supported for caches with geo-replication, or Flash Optimized tier.
Choose a representative application scenario and define success criteria before testing. Assess application behavior, integration effort, and performance against your requirements, then use those findings to inform your team’s adoption decision.
Explore Azure Managed Redis today
For creating a new Azure Managed Redis cache with Redis 8.2, sign up for the preview now!