Has HTTP/2 performance improved as of Go 1.24? Last I checked forcing HTTP/1.1 everywhere was a massive improvement in throughout and latency for a very busy distributed system.
It has a bit, but there are fundamental issues. Given the way the Go runtime wants to deal with reading and writing sockets, HTTP/2 requires 2 extra goroutines per connection, and the bouncing around over channels that this implies. This might not be a law of physics but there isn't another obvious way to do it in Go.
In the past when I wanted a really fast Go service using HTTP/2 I put the HTTP server in a C++ subprocess that handled the sockets and communicated with the Go application over a pipe. That was nice and fast, avoided the congestive collapse that Go suffers with too many runnable goroutines.
Note that AWS ALB does not support h2c. When the client and server do, ALB will dutifully forward the h2c header and fail to handle the upgraded response.
Except you can do http1 requests in bash easily and many http3 libraries are bad anyway so you don't get the advantages of using it.
I wanted to do something fast using http3 but it ended up being way faster using ad-hoc code I wrote using http1. It would be even faster if I did it with http3 but hand writing for that protocol is a nightmare so here we are…
I love that anyone can write a blog post like this that will get slurped into all the models and we can just say: "use terraform to deploy H2C on GCR"... and it will know exactly what to do.
To use a Analogy, AI is acting like pip install library in this case. Do you really want to do performative acts of Sisyphus or are more interested in end result?
This is excellent news for human persons. Protocol implementations that only allow TLS are not very robust without human maintenence for more than a few years. That said, the human person use cases for HTTP/2 are pretty limited. Generally HTTP/1.1 is a better choice.
https://github.com/rclone/rclone/commit/ad8a108453f3ce983fb6...
It is interesting to dig into why.
There was a security vulnerability in golang.org/x/net/http2/h2c which meant govulncheck warned about it in the CI.
So I updated it and got a warning from the linter that the h2c sub package was deprecated in the latest version, so I removed it.
That is a lot of great tooling working to make things more secure in the Go ecosystem.
It does make work for maintainers though, and the Cambrian explosion of AI discovered security vulnerabilities has been particularly trying!
In the past when I wanted a really fast Go service using HTTP/2 I put the HTTP server in a C++ subprocess that handled the sockets and communicated with the Go application over a pipe. That was nice and fast, avoided the congestive collapse that Go suffers with too many runnable goroutines.
I wanted to do something fast using http3 but it ended up being way faster using ad-hoc code I wrote using http1. It would be even faster if I did it with http3 but hand writing for that protocol is a nightmare so here we are…