Nobody wants a cache. What people want is for reads to be fast, and for that speed to stop costing them.
When reads get expensive, most teams have two options: scale up with bigger instances and read replicas, or build caching by hand and pay engineers to maintain it. Both can be serious spend, and both hold only until the next time traffic shifts.
Today we're launching the Readyset Platform. Caching now runs in your infrastructure and tunes itself as workloads change. It decides what's worth caching, adjusts freshness per query, and directs memory where it delivers the most return.
The result is smaller instances, fewer read replicas, and engineering time back on your roadmap instead of your cache. It's the cache you can leave on, and it will be better four months from now than it is today, at no additional cost to you.
Our CEO Gautam Gopinadhan wrote up the thinking behind it, where automatic caching goes wrong, and what we built to close those gaps.
Top comments (0)