Week 909

Week beginning Monday, 15th June 2026.

We’ve continued to work on performance-related improvements for the Mavis project for NHS Digital. The service is expecting an increased amount of traffic and data volumes and it’s important that we give everyone confidence that the system will continue to work as intended.

We have a performance environment available to do some load tests but replicating exactly the traffic patterns we’d expect in production is challenging. We’ve mostly been convincing ourselves that various parts of the service will function correctly under increased traffic if we allocate additional compute resources to them. It feels like we’re getting to a good place with that - it’s a much easier story to tell if we can show a linear relationship between traffic and the amount of compute we need. We can come back to making performance optimisations when we’re under a little less time pressure.

On Tuesday Chris R gave a great demo of the Playwright trace viewer and how we can use it to work out why certain end-to-end tests are flaky. Coincidentally our Ruby friends at Thoughtbot put a blog post out today on this topic too.

The team have also shown some interest in how we approach pull requests and “telling stories” with our git commit history. So we’re planning to dust off a workshop we gave on this while working with the Raspberry Pi Foundation a couple of years ago.

On Wednesday we took a day away from client work to get together in Cambridge. We had a good lunch and walk and then spent a couple of hours watching the May Bumps from the garden of The Plough in Fen Ditton.

I’ve ended the week catching up on some admin and conversations generated by last week’s Scaling Co-operation event. I’m packing up soon to take my kids for a weekend camping in the New Forest. I’ll be at Brighton Ruby next Thursday, do stop me for a chat if you see me!