# Plug POST performance

**URL:** <https://forum.elixirforum.com/t/plug-post-performance/473>\
**Category:** Questions\
**Tags:** performance\
**Created:** [April 30, 2016, 2:31pm UTC](https://forum.elixirforum.com/t/plug-post-performance/473 "2016-04-30T14:31:23Z")\
**Posts on this page:** 1\
**Showing post:** 19

<div class="post-metadata">

**Author:** ![apenney](https://forum.elixirforum.com/user_avatar/forum.elixirforum.com/apenney/32/1292_2.png) [@apenney](https://forum.elixirforum.com/u/apenney)\
**Post date:** [May 6, 2016, 4:54pm UTC](https://forum.elixirforum.com/t/plug-post-performance/473/19 "2016-05-06T16:54:26Z")

</div>

I spent some time last night “tinkering” with interesting results.

Cowboy 2.0, using a hacked up copy of their examples/post\_data can do 106k/sec 204’ing without reading the body on the machines I struggled to get over 25k on with elixir/plug.

Elli could do 160k on the same hardware.

I tried to use one of the elli plug adapters and got 2k QPS through Elixir. The weirdest part of my Elli testing was I was using `rebar3 shell` and that creates +A 0 and +K false and I still got 160k. I am sure if I build a release and tune it properly I can drive that upwards of 250-300k.

Just figured I’d put a mini update as I’m not finished (very slowly) trying to untangle this performance issue. My next plan is cowboy 1.1, which is what Plug is using, to see if I reach a bottleneck there.

Then adding raw erlang body reading into both cowboy and elli to see what that looks like.

---

_[View the full topic](https://forum.elixirforum.com/t/plug-post-performance/473)._