If you’ve ever stretched a cache across more than one region, you know the trade-off. You pick a primary, point writes at it, and replicate outward. Readers far from the primary get stale data or pay the round trip. When the primary’s region has a bad day, so does everyone else.

Swytch was built to remove that trade-off. It’s a distributed cache with no leader. Every node in every region accepts reads and writes, and the cluster keeps the data consistent across all of them. There’s no primary to fail over and no replica that’s stale.

What “leaderless” means in practice

In a leader-based setup, one node decides the order of writes. That’s simple, but it makes geography your problem. In Swytch, nodes coordinate among themselves. A write in Frankfurt and a read in San Francisco see the same value without either one routing through a third location.

It also changes what failure looks like. Losing a node or a region doesn’t trigger an election or a promotion step. The remaining nodes keep serving.

Copying every key to every region is how multi-region caches get expensive. Swytch replicates based on interest instead. Data moves to the regions that actually ask for it, so you’re not paying to store and ship keys nobody in that region reads. You still get one globally consistent view, without a full copy everywhere.

It speaks Redis

Swytch is Redis-wire compatible. Your existing clients, libraries, and tooling work unchanged. You don’t need a new SDK, and you don’t rewrite your data access layer. For most teams, trying it means changing a connection string.

Open source first

The core is open source, and it’s already out in the wild. Over 300 nodes are running it today, spread from Japan to California to Sweden to Malaysia. If you want to run it yourself on Kubernetes or bare metal, you can.

∙ Teams with users in more than one region who are tired of choosing between stale reads and slow writes.

∙ Anyone running a primary/replica Redis setup who has been burned by failover.

∙ Platform and edge teams who want the cache to live close to the workload without building their own consistency layer.

Try it without installing anything

The fastest way to see it is a managed trial cluster. Sign in at getswytch.com/login, grab your endpoint, and point redis-cli at it:

redis-cli -h <your-endpoint> -p <port>

SET hello world

GET hello

That’s a live multi-node cluster answering in a Redis dialect you already know. From there you can connect your own app and watch how it behaves under your traffic.

Questions, rough edges, and feature requests go to the community Discord. We read everything there.

Where we’re headed

Swytch B.V. is based in Utrecht, and the cache is one piece of a broader data layer we’re building. If you run infrastructure across regions and want to compare notes, we’d like to hear what’s breaking for you.