Hello, I’m new in whole elixir world and I’m going to write a master degree about High user traffic in webapps. Basicaly its going to be comparison rails server way to elixir/phoenix way.
I would be very pleased if you could recommend me some stuff to read about. For now I found a cowboy documentation, and I know that its spawning process for each request. But its way not enough of knowledge and materials. What should I read and where to look for informations ?
1 Like
Also look at:
1 Like
This is a good one too https://www.bignerdranch.com/blog/elixir-and-io-lists-part-2-io-lists-in-phoenix/
2 Likes
Thank you very much ! This is very helpful for writing app.
Could you point me a way to find materials about processing request in phoenix ?
I think its erlangs vm job, right ?
1 Like
Talking about IOlists, this one is also very insightful: http://www.evanmiller.org/elixir-ram-and-the-template-of-doom.html
2 Likes
Phoenix builds on the Elixir lib Plug, which uses the cowboy webserver, which is written in Erlang. (in theory, Plug is webserver-agnostic but so far there is only one adapter, which is for cowboy, which has become the de-facto webserver for Erlang/Elixir-based projects).
Cowboy is well-documented, its user guide is found here, including a section on its internals. For the webserver, Cowboy itself uses Ranch, a TCP socket acceptor pool which (iirc) does the actual job of spawning one process per connection. It has a user guide as well.
Another Erlang-based webserver is Elli, which I think people also have written Plug adapters for, but Elli is more barebone than Cowboy and not as well-maintained.
2 Likes
“Runtime scheduling: theory and reality” by Eben Freeman:
2 Likes
