Valkey: The Open-Source Redis Alternative After the License Change

In March 2024, Redis Ltd. changed the license for Redis from the permissive BSD license to the Server Side Public License (SSPL) and Redis Source Available License (RSAL), effectively ending Redis's status as true open-source software. Within weeks, a coalition of cloud providers including AWS, Google Cloud, and Alibaba Cloud forked Redis 7.2 and donated the project to the Linux Foundation under the name Valkey. If you run Redis in production today, this change affects your licensing exposure, your upgrade path, and — depending on your workload — potentially your performance ceiling.

This guide covers what the license change actually means, how Valkey differs from Redis today, what a migration looks like in practice, and where the managed cloud offerings stand as of early 2026.

What Happened with Redis Licensing

Redis was governed by the three-clause BSD license from its creation in 2009 through version 7.2 (released September 2023). That license allowed anyone to use, modify, and distribute Redis freely — including cloud providers who sold managed Redis services without contributing code back upstream.

On March 20, 2024, Redis Ltd. announced that all Redis releases from version 7.4 onward would be dual-licensed under the Server Side Public License (SSPL) v1 and the Redis Source Available License (RSAL) v2. Neither license is approved by the Open Source Initiative (OSI). The SSPL, originally written by MongoDB, requires that any service offering the software as a network service must open-source the entire service stack — a condition that cloud providers cannot reasonably satisfy and that many legal teams treat as commercially prohibitive.

The practical upshot: teams that need a genuinely open-source, production-grade, in-memory data store now have a clear migration target in Valkey, and a growing set of managed offerings to back it.

What Is Valkey

Valkey is an open-source, high-performance key-value store forked from Redis 7.2.4 and donated to the Linux Foundation in March 2024. The founding contributors include AWS, Google Cloud, Alibaba Cloud, Oracle, Ericsson, and several independent maintainers. The project moved fast: Valkey 7.2.5 shipped within weeks of the fork announcement, and Valkey 8.0 (a major feature release) arrived in September 2024.

Key governance facts:

• License: BSD 3-Clause (same as Redis 7.2 and earlier)
• Governance: Linux Foundation project with a Technical Steering Committee (TSC)
• Versioning: Valkey maintains the 7.2.x branch for stability and advances the 8.x series for new features
• Repository: github.com/valkey-io/valkey

Valkey 8.0 introduced the most significant architectural change: improved multi-threaded I/O. Redis has long been single-threaded for command execution (with background threads for persistence and replication). Valkey 8.0 extends the threading model to allow multiple I/O threads to handle network reads and writes concurrently, which reduces latency under high connection counts and increases aggregate throughput on machines with 4+ cores.

API and Protocol Compatibility

Valkey is fully compatible with the Redis Serialization Protocol (RESP2 and RESP3) and implements the complete Redis 7.2 command set. This means:

• All Redis clients work with Valkey without modification. This includes redis-py, ioredis, Jedis, Lettuce, go-redis, StackExchange.Redis, and every other mainstream client library. The clients speak RESP; Valkey speaks RESP.
• RDB persistence files from Redis 7.2 can be loaded by Valkey and vice versa.
• AOF (Append-Only File) format is compatible across the two at the 7.2 baseline.
• Cluster topology, replication, and Sentinel configuration formats are unchanged.

The one compatibility boundary to be aware of: Modules. Redis modules compiled against the Redis module API may need recompilation or minor adjustments for Valkey. Most popular modules (RedisJSON, RediSearch, RedisBloom) have Valkey-compatible builds or are being ported by the community. Check the Valkey module compatibility matrix before migrating if you depend on modules.

Migration Steps

For the majority of deployments that use Redis as a cache or session store without modules, migration is straightforward.

Self-Managed Migration (Standalone or Sentinel)

1. Install Valkey on the target server. Packages are available for Debian, Ubuntu, RHEL, and Alpine, and official Docker images are published to docker.io/valkey/valkey.
2. Copy your redis.conf to valkey.conf. Nearly all directives are identical. Valkey will warn on startup about any unrecognised directives.
3. Take an RDB snapshot on the source Redis instance (BGSAVE), then transfer the dump.rdb to the Valkey data directory.
4. Start Valkey and verify it loads the RDB cleanly. Check INFO persistence for rdb_last_bgsave_status:ok.
5. Update your application connection string. Change the host/port to the Valkey instance. No other changes are required.
6. Monitor for the first 24 hours: check hit rate, latency percentiles, and replication lag if applicable.

Self-Managed Migration (Cluster Mode)

Cluster migrations require a rolling replacement. The recommended approach is to use CLUSTER MEET to introduce Valkey nodes into an existing Redis cluster, then CLUSTER FAILOVER each shard one at a time. Redis 7.2 and Valkey 7.2 nodes can participate in the same cluster during the transition window because the cluster bus protocol is identical.

# On each Valkey node being introduced to the cluster
valkey-cli -h  -p 6379 CLUSTER MEET  6379

# Verify cluster sees the new node
valkey-cli -h  CLUSTER NODES

# Trigger a failover on a replica that has been promoted to Valkey
valkey-cli -h  -p 6379 CLUSTER FAILOVER

Client Library Configuration

No client changes are required for basic compatibility. If you want to use Valkey-specific features introduced after 7.2, install the official Valkey client for your language:

# Python
pip install valkey

# Node.js
npm install iovalkey

# Go
go get github.com/valkey-io/valkey-go

These clients are wire-compatible with Redis 7.2 as well, making them safe to adopt before the migration completes.

Performance

Benchmarks published by the Valkey project and independently by AWS show meaningful throughput gains in Valkey 8.x versus Redis 7.2 on multi-core instances, primarily because of the improved I/O threading.

Representative results from AWS internal benchmarks (r7g.2xlarge, 8 vCPUs, 100k connections, mixed GET/SET workload):

• Redis 7.2: ~450,000 ops/sec
• Valkey 7.2: ~455,000 ops/sec (comparable — same codebase)
• Valkey 8.0 (2 I/O threads): ~680,000 ops/sec
• Valkey 8.0 (4 I/O threads): ~850,000 ops/sec

Valkey's I/O thread count is controlled by two configuration directives:

# valkey.conf
io-threads 4
io-threads-do-reads yes

A general recommendation: set io-threads to half the number of available vCPUs, and always benchmark before and after enabling io-threads-do-reads, as it can occasionally increase latency variance on memory-bound workloads.

Managed Options

The major cloud providers have moved quickly to offer Valkey as a managed service.

AWS ElastiCache for Valkey

AWS announced ElastiCache support for Valkey in August 2024, with general availability of Valkey 7.2 and 8.0 engine versions. Existing ElastiCache Redis clusters can be migrated to Valkey in-place using a blue/green deployment — no data export required. ElastiCache for Valkey retains the same parameter groups, security groups, and VPC integration as the Redis engine.

AWS MemoryDB for Valkey

MemoryDB (AWS's durable, Redis-compatible service with multi-AZ transactional logs) added Valkey engine support in late 2024. MemoryDB for Valkey is the appropriate choice when you need Redis-level APIs with durable storage guarantees — it persists every write to a multi-AZ transaction log before acknowledging the client.

Other Providers

Google Cloud Memorystore added Valkey support in 2024. Aiven, Upstash, and several other DBaaS platforms have followed. For self-hosted deployments, Bitnami Helm charts and the official Valkey Docker image are the fastest path to a production-ready cluster on Kubernetes.

Comparison: Redis vs Valkey vs KeyDB vs Garnet

KeyDB (acquired by Snap Inc. in 2022) offers true multi-master replication and has been multi-threaded since its inception, but its development velocity has slowed and its governance is less transparent than Valkey's Linux Foundation model. Garnet is Microsoft Research's reimplementation of the Redis protocol in C#/.NET, targeting Windows Server and Azure scenarios; it is not a Redis fork and its compatibility surface is narrower. For most teams migrating from Redis, Valkey is the straightforward choice.

How JusDB Can Help

Evaluating the right in-memory data store for your production workload — whether that is self-managed Valkey, AWS MemoryDB for Valkey, or a different architecture — involves tradeoffs across performance, durability, cost, and operational overhead. JusDB covers the managed database landscape in depth, from configuration tuning guides to architecture decision frameworks for teams choosing between caching layers and persistent stores.

Browse the Valkey database page for a full breakdown of Valkey's feature set, a comparison of managed providers, and links to official documentation. If you are evaluating multiple databases for a new project, the JusDB comparison tool lets you put Valkey, Redis, KeyDB, and Garnet side by side across the dimensions that matter for your workload.