// the find
olric-data/olric
Distributed, in-memory key/value store and cache. It can be used as an embedded Go library and a language-independent service.
Olric is a distributed in-memory key/value store written in Go that runs either embedded inside your application or as a standalone server. It uses Redis Serialization Protocol, so any Redis client speaks to it out of the box. The target user is a Go developer who wants clustered shared memory without standing up Redis and its associated infrastructure.
The dual-mode deployment (embedded library vs. standalone server) is genuinely useful — you can start with embedded for simplicity and migrate to client-server without changing your application logic. Using RESP means you don't need a custom client in any language; redis-cli works on day one. The storage engine (internal/ramblock) is GC-friendly by design, packing values into pre-allocated byte slices rather than exposing them as Go heap objects — this matters a lot for latency at high cardinality. Consistent hashing with automatic rebalancing on node join/leave works without an external coordinator like ZooKeeper or etcd, which is a meaningful operational simplification.
The consistency model is PA/EC (available over consistent under partition), and the README is honest about this but the locking primitive is still labeled 'approximate' and 'for non-critical purposes' — which makes it nearly useless for any real distributed coordination need. Authentication is password-only with no TLS built in, so you're trusting your network overlay for any security; that's a non-starter for multi-tenant or cloud deployments without extra infrastructure. The module rename from buraksezer/olric to olric-data/olric happened at v0.6 and the warning is still prominent in the README, suggesting adoption is fragmented and tooling that pinned the old path will silently break on updates. At 3.4k stars and 147 forks after years of development, the community is thin — the Discord likely has a small number of active members, and the sponsorship note implies the maintainer bandwidth is constrained.