Redis vs Valkey fork comparison

Redis 8 vs Valkey: The Fork That Split the Caching World

Redis Labs changed Redis to a dual license (RSALv2 + SSPLv1) that prevents cloud providers from offering managed Redis services. The Linux Foundation forked it as Valkey under BSD license. Redis 8 vs Valkey — which should you choose? This guide cuts through the licensing drama to focus on what matters: technical differences, migration effort, and long-term risk.

What Actually Happened

In March 2024, Redis Ltd. (the company, formerly Redis Labs) changed Redis’s license from BSD to a dual RSALv2/SSPL license. This means cloud providers (AWS, GCP, Azure) can no longer offer managed Redis services using Redis source code. Additionally, any company that embeds Redis into a product they sell as a service is affected.

In response, the Linux Foundation launched Valkey — a fork of Redis 7.2.4 (the last BSD-licensed version) — with backing from AWS, Google, Oracle, Ericsson, and Snap. Valkey is fully open-source under BSD license and aims to be a drop-in replacement for Redis.

For you as a developer or operator, the impact depends on how you use Redis:

Self-hosted Redis for your own apps: The license change doesn’t affect you. You can use Redis 8 freely for your own applications.
Managed Redis service (ElastiCache, Memorystore): AWS has already switched ElastiCache to Valkey. GCP Memorystore is transitioning. Your managed Redis service may already be Valkey under the hood.
Embedding Redis in a product you sell: You need to evaluate the RSAL/SSPL restrictions. Valkey is the safe choice.

Technical Comparison: Redis 8 vs Valkey

Both projects started from the same codebase. Here’s where they’ve diverged:

COMPATIBILITY (as of March 2026):
  Redis protocol:     Both fully compatible (RESP2 + RESP3)
  Redis commands:     99.9% compatible (Valkey added some, Redis added some)
  Redis modules:      Redis Stack modules are NOT available in Valkey
  Redis Sentinel:     Both supported
  Redis Cluster:      Both supported
  Client libraries:   All Redis clients work with Valkey (same protocol)

REDIS 8 EXCLUSIVE FEATURES:
  - Redis Search (full-text search built-in)
  - Redis JSON (native JSON document type)
  - Redis Time Series
  - Redis Bloom (probabilistic data structures)
  - Redis Graph (deprecated but still available)
  These are "Redis Stack" modules — proprietary, not open-source

VALKEY EXCLUSIVE FEATURES:
  - Multi-threaded I/O (significantly faster for high-connection-count scenarios)
  - Improved memory efficiency (RDMA support for cluster replication)
  - Enhanced slot migration for Cluster mode
  - Community-driven development with faster release cycles

PERFORMANCE:
  Single-thread operations: Identical (same core engine)
  Multi-connection throughput: Valkey wins with multi-threaded I/O
  Memory efficiency: Comparable (Valkey slightly better in cluster mode)
  Latency p99: Identical for standard operations

Redis vs Valkey performance comparison — Core performance is identical — differences emerge in modules and multi-threaded I/O

Migration from Redis to Valkey

For standard Redis usage (caching, sessions, pub/sub, sorted sets), migration is straightforward — Valkey is wire-compatible with Redis. Your existing client libraries (redis-py, ioredis, Jedis, go-redis) work without code changes. Just point them at the Valkey endpoint.

# Your existing Redis code works with Valkey unchanged
import redis

# Before: Redis
# client = redis.Redis(host='redis.internal', port=6379)

# After: Valkey — same client library, different host
client = redis.Redis(host='valkey.internal', port=6379)

# All operations work identically
client.set('user:1001:session', session_data, ex=3600)
client.hset('product:xyz', mapping={'name': 'Laptop', 'price': '999.99'})
client.zadd('leaderboard', {'player1': 100, 'player2': 95})
result = client.zrangebyscore('leaderboard', 90, 100)

If you use Redis Stack modules (Search, JSON, TimeSeries), migration is harder. Valkey doesn’t include these modules. You’ll need to either: keep using Redis for these features, migrate to alternative solutions (Elasticsearch for search, PostgreSQL JSONB for JSON documents, TimescaleDB for time series), or wait for community alternatives to mature.

Which Should You Choose?

Choose Valkey when: You use a managed cloud service (it may already be Valkey), you want fully open-source with no license risk, you don’t use Redis Stack modules, or you’re starting a new project and want the safest long-term choice.

Choose Redis 8 when: You need Redis Stack modules (Search, JSON, TimeSeries) and don’t want to use separate tools for these features, you’re self-hosting for your own applications (the license doesn’t restrict this), or you have Redis Enterprise support contracts.

Caching architecture decision — For standard caching: Valkey. For Redis Search/JSON/TimeSeries: Redis 8. For new projects: Valkey.

The Long-Term Outlook

Valkey has momentum: AWS, Google, and Oracle are backing it. Major cloud providers are migrating their managed offerings. The contributor community is growing rapidly. Redis has a strong product (Redis Stack) but faces increasing isolation as the ecosystem moves to Valkey.

For most teams, the practical advice is: use Valkey for new deployments, migrate existing Redis to Valkey when convenient, and only stick with Redis if you have a hard dependency on Redis Stack modules.

Related Reading:

Resources:

In conclusion, Redis 8 vs Valkey is primarily a licensing decision. Technically, they’re nearly identical for standard use cases. Choose Valkey for open-source safety and cloud compatibility. Choose Redis 8 if you need proprietary Redis Stack modules. Either way, your application code stays the same.