https://github.com/probably-not/regbench
Benchmarks of registries in Elixir in order to understand what issues there are in scale.
https://github.com/probably-not/regbench
Last synced: 2 months ago
JSON representation
Benchmarks of registries in Elixir in order to understand what issues there are in scale.
- Host: GitHub
- URL: https://github.com/probably-not/regbench
- Owner: probably-not
- Created: 2024-02-28T20:37:32.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2024-05-13T13:14:58.000Z (about 2 years ago)
- Last Synced: 2025-01-07T06:32:14.116Z (over 1 year ago)
- Language: Elixir
- Size: 67.4 KB
- Stars: 3
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Regbench
Some benchmarks of registries in Elixir in order to understand what issues there are in scale.
Benchmarks were ported (or at least I attempted to port them) from the [@ostinelli blog post about Process Registries in Erlang](http://www.ostinelli.net/an-evaluation-of-erlang-global-process-registries-meet-syn/).
Of course... I'm not great at Erlang, so maybe I screwed up the porting...
## Why?
I personally don't really believe people need to jump on the "built-in Erlang `:global` is bad at scale!!!" train that everyone immediately jumps on when people start talking about distributed registries in Elixir. I don't think anybody who is jumping on that train is actually getting to a point where it seriously affects scale (seeing as the `:global` can get to a few thousand registrations per second according to the blog post and lookups are consistently very low). However, the post was written in 2015, and I wanted to see whether Syn (and the conclusions that [@ostinelli](https://github.com/ostinelli) came to) still hold up.
## Results
TODO: Fill in results!