GitHub
Getting startedInstall, connect and make your first request.ConfigurationServer, memory, persistence and environment settings.Clients & SDKsConnect with Go, Node.js, Python or the Redis CLI.Data structuresStrings, hashes, lists, sets, streams and more.Architecture256 shards, RESP networking and the storage engine.PersistenceSnapshots, append-only files and recovery.Security & ACLAuthentication, TLS and scoped command access.Clustering & replicationGossip, hash slots, replicas and Sentinel.MonitoringPrometheus, slow logs and health checks.HTTP APIIntegrate using JSON over HTTP.Command referenceSearch command syntax across the upstream modules.BenchmarksStorage and TCP results, hardware and methodology.FeaturesEverything inside the CacheStorm binary.ChangelogReleases and development changes.Brand assetsDownload the CacheStorm SVG logo and mark.
↑ ↓ to navigate · Enter to openCacheStorm docs

FAST. WITH CONTEXT.

Performance you
can look into.

Numbers tell a better story when you know what they measure. Here are the upstream results, their scope and a starting point for your own benchmarks.

Context is part of the result.

These are repository-reported measurements from the pinned upstream README, not fresh benchmarks run for this website. Hardware: AMD Ryzen 9 9950X3D, 64 GB RAM, Windows 11. In-memory results exclude network transport; end-to-end results include the TCP round trip. Persistence, client behavior, key sizes, contention and hardware can change the outcome.

Read the original results ↗

01 / INSIDE THE ENGINE

Direct storage operations.

In-memory store benchmarks. No TCP connection, protocol parsing or network round trip.

OperationOperations / secondTime / operationAllocations / operation
GET · sequential20M50 ns0
GET · parallel, 32 logical processors106M9.4 ns0
SET · sequential1.3M796 ns3
SET · parallel, 32 logical processors2.2M449 ns3
Tag count73M13.7 ns0
Tag invalidation · 10K keys5101.96 ms—

02 / THE COMPLETE CONNECTION

End-to-end TCP operations.

A separate workload including the full TCP round trip. These results describe a different measurement from direct storage calls.

OperationOperations / secondReported time / operation
SET32K37 µs
GET33K37 µs
HSET33K38 µs
ZADD31K36 µs
XADD31K39 µs

Values are reproduced as reported upstream, including rounding. Operations per second and listed timings are not always exact reciprocals. Treat them as upstream observations and reproduce your own workload.

03 / MAKE IT YOUR MEASUREMENT

Your workload is
the benchmark that matters.

Run the benchmark suite on the version you plan to deploy. Record the commit, Go version, CPU, OS, dataset size, concurrency and whether persistence is enabled.

Then exercise the complete application path with your selected client, real command mix and representative data. Measure tail latency as well as throughput.

Explore benchmark source ↗
run your own benchmarks
git clone https://github.com/CacheStorm/CacheStorm.git
cd CacheStorm
git checkout baf028b1276b411641edd92e7135202fd114f4e1

# Record the environment
go version

# Storage benchmarks
go test ./internal/store/... -run=^$ -bench=. -benchmem

# Additional benchmark workloads
go test ./benchmarks/... -run=^$ -bench=. -benchmem