[← Blog](https://watasu.io/blog) April 27, 2026 # Valkey: give your app a working memory The early-access shelf is one item lighter: **Valkey** is now open to every Watasu account. It’s the piece almost every production app reaches for sooner or later — the fast, in-memory layer that sits next to your real database and absorbs everything that shouldn’t hit it. ## What Valkey is Valkey is an open-source, in-memory data store that **speaks the Redis protocol** — your existing client libraries, frameworks, and muscle memory work unchanged. We even inject the connection URL as `REDIS_URL` on purpose, so the ecosystem’s defaults “just work” without a config hunt. The mental model is simple: PostgreSQL is your system of record; Valkey is your working memory. Reads in microseconds, writes that don’t queue behind disk, and data structures (lists, sets, sorted sets, TTLs) designed for the patterns below: - **Caching** — computed values and query results that are expensive to make and cheap to throw away - **Sessions** — fast lookups on every request, with TTLs doing the cleanup - **Rate limiting** — counters with expiry, the canonical implementation - **Ephemeral coordination** — locks and leader election with TTL safety nets - **Queue acceleration** — hot-path state for your background jobs ## One command ``` watasu addons:create valkey --app my-app ``` Connection details land in your app’s config vars, and an interactive session is always one command away: ``` watasu valkey:cli --app my-app ``` (The `redis:cli` alias works too — we know about muscle memory.) ## Pick a size, not a topology Plans are deliberately simple. A **€1 Hobby instance** — single-node, non-persistent — covers prototypes, review apps, and throwaway caches. For everything real, **Premium starts at €5/mo**, and every Premium size ships the same production posture: **AOF persistence, a managed HA standby, and automatic failover**. You never choose between “the cheap one that loses data” and “the safe one you can’t afford yet” — you just pick how much memory and how many connections you need, from 50 MB up to 100 GB and 65,000 connections. Persistent plans support backups — manual everywhere, scheduled on the higher sizes — with the same non-destructive restore-and-promote workflow as every Watasu database. One honest note for latency tuners: Valkey’s value here is failover topology and durability, not exotic storage. If you’re benchmarking AOF write latency at the edge of what’s possible, read the [storage notes in the docs](https://docs.watasu.io/addons/valkey/) before you size your plan. ## Get started Attach one to your app, point your cache or session store at `REDIS_URL`, and watch your database breathe easier. The [Valkey docs](https://docs.watasu.io/addons/valkey/) cover plans and backups, and the [pricing page](https://watasu.io/pricing#valkey) has the full ladder. Questions? — we read everything.