Fastly uses the H2O reverse proxy for fast and secure TLS termination over QUIC, HTTP/3, HTTP/2, and 1.1.
The project site compares its benchmarks only to Nginx. I'd love to see a more recent comparison that includes ATS (Apache Traffic Server), HAProxy, and Varnish/Hitch as fellow reverse proxies for TLS termination.
"Putting up the appropriate guardrails to protect ourselves from regressions – then pairing that with a trail of breadcrumbs so that we can dive in and quickly identify what the source is when a regression occurs – are essential steps to ensure that when we make our sites fast, they stay that way."
Awesome post by @tkadlec, filled with high-level wisdom and ground-level best practices.
Keeping your site fast is a crucial and endless game. Yet it's perilously easy to lose focus and suffer from regressions. In this excellent, detailed, best-practice-filled post, @tkadlec uses the analogy of guardrails (automated testing & alerts) and breadcrumbs (deployment tracking) to make performance more visible throughout the dev process.
🔹 Improvements to the Speculation Rules API by @tunetheweb
🔹 Shared compression dictionaries by @malchata
🔹 The importance of web performance to information equity by @Schouten_B
Chris rounds up great stuff by @scottjehl@csswizardry Rick Viscomi & Tim Severien (and was kind enough to include my psychology of speed post). I also learned about the existence of the 1MB club!
I'll be keeping a keen eye on RUMvision over the next few days and share some insights – I've pushed Shopify hard on #webperf so it's exciting to see how it translates to the real world!
These days there's a lot of talk about what INP is, and not enough talk about how to fix it. There's also a tendency to oversimplify INP, when it's not a simple metric. This must-read post by @andydavies tackles the hard stuff.
The next edition of Speed Matters is dedicated to INP: what it is, how to track it, how to validate it against your business/UX metrics, and – most important – how to improve it.