Re: [dev] [quark] Performance issues

From: <maillists.rulmer_AT_mailbox.org>
Date: Tue, 24 Sep 2019 19:58:01 +0200

Hi Laslo and Anselm,

Laslo Hunhold <dev_AT_frign.de> wrote:
> On Mon, 23 Sep 2019 13:07:47 -0700
> Anselm Garbe <garbeam_AT_gmail.com> wrote:
> > Thanks for doing that and letting me know. So the reason you see this
> > performance penalty in contrast to the other web servers you mention
> > is, that quark is a fork() based web server (and current HEAD is still
> > fork() based). That's the whole reason ;)
>
> I also guessed that this might be the reason before the results came.
> Now it is pretty certain, especially when we also note that fork() has
> a higher penalty on OpenBSD compared to Linux.

Thanks for your assessments! I think you've hit the nail on the head. I
got my initial test results on an OpenBSD machine. I tested the golang
server and quark again on a Linux machine today and the difference is
much less pronounced:

quark (Linux): 14494 requests/second
golang (Linux): 16601 requests/second

As you'd expect the difference also declines when the requested file is
larger. With an index.html, that is 1MiB in size I got these results:

quark (OpenBSD; 1MiB index.html): 160 requests/second
golang (OpenBSD; 1MiB index.html): 198 requests/second

> Quark is actually very lean and offers 99% of the features you would
> expect for a static server. I personally am a big fan of OpenBSD's
> httpd and will use it on the server I am currently setting up.
>
> I see quark's role more like a "drop-in" server you can invoke
> literally in seconds from the command line to share some data on the
> network. In this form, it is not so trivial with other servers. For
> "settled" servers with a fixed configuration, OpenBSD httpd works
> flawlessly in my opinion!

Thanks for the insight! I was thinking about using quark instead of
OpenBSD's httpd, because it simpler to use and probably even more
secure. I think ~1000 requests/second is still plenty for all my
projects, but in case I need the extra performance some day, I know what
the bottleneck is.

With best regards,
Richard Ulmer
Received on Tue Sep 24 2019 - 19:58:01 CEST