Hello everyone,

Builds on Railway are going to get a lot of love, well, they are now because they have to get love.

We're growing quickly which means that our old build infrastructure has been struggling to keep up. We got the order from on high to drop everything else and get build times to be as quickly as possible. Many of you are shipping code for work and for fun and the time it take to see your image build is maybe the most magical experiences, if it's good, and not so magical if it's bad. Namely, no one should have to wait.

So we're going to be taking a multi-pronged approach to making sure we really close the gap between the time you press enter on railway up and when your application goes live.

1. We've now added significantly more telemetry on avg. build times and outliers for builds. When these times spike, we get paged, simple as. We'd rather lose sleep than know the experience is not stellar.
2. Our infra team is looking to optimize the co-location of build resources to make it so that network hops or scheduling on a host in the fleet aren't the limiting reagent.

3. For Railpack, we're looking into investing where the bottlenecks are in terms of package downloads or package installation on our machines.

We will need some help, if there are workloads that you see on Railpack take a long time in your build, please call it out to us so we can optimize it. Example, if you are noticing an npm install or a uv pip install take longer than expected. Please flag it on this thread or a support request.

This season, we're thankful for you, your business, and your support while we scale. Happier shipping everyone.