Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 16 Feb 2003 21:47:44 -0700 (MST)
From:      Alex Rousskov <rousskov@measurement-factory.com>
To:        Pawel Jakub Dawidek <nick@garage.freebsd.pl>
Cc:        Scott Long <scott_long@btc.adaptec.com>, Sam Leffler <sam@errno.com>, Brad Knowles <brad.knowles@skynet.be>, freebsd-current@freebsd.org
Subject:   Re: 5-STABLE Roadmap
Message-ID:  <Pine.BSF.4.53.0302162130370.46493@measurement-factory.com>
In-Reply-To: <20030216214322.GB10767@garage.freebsd.pl>
References:  <20030216184257.GZ10767@garage.freebsd.pl> <3E4FFDD3.9050802@btc.adaptec.com> <20030216214322.GB10767@garage.freebsd.pl>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 16 Feb 2003, Pawel Jakub Dawidek wrote:

> On Sun, Feb 16, 2003 at 02:08:35PM -0700, Scott Long wrote:
> +> Pawel Jakub Dawidek wrote:
> +>
> +> >On Thu, Feb 13, 2003 at 08:28:43PM -0800, Sam Leffler wrote:
> +> >+> This can quickly turn into a bikeshed, but suggest ones.  We're
> +> >looking for
> +> >+> good benchmarks. [...]
> +> >
> +> >Look at:
> +> >
> +> >	http://www.web-polygraph.org
> +> >
> +> >It provides tests for www-cache/proxy stuff.
> +> >We can test many things with it:
> +> >
> +> >	- how fast could we generate workload,
> +> >	- how heavy load could we handle,
> +> >	- how fast is squid running on FreeBSD,
> +> >	- how fast is squid rewritten with libkse,
> +> >	- etc.
> +> >
> +> >And this is good stablility test.
> +> >This is real good and free stuff, I use it on 4.x.
> +> >
> +> Thanks for the pointer, this looks very interesting.  How hard
> +> is it to set up?  [...]
>
> Setting it up is quite simple, but it doesn't compile with gcc 3.x...

There were too many non-backwards compatible changes in 3.x and the
compiler was not stable enough last time we checked. We will probably
check and port again some time soon. GCC 2.9x should work fine though.

Polygraph is relatively easy to setup on FreeBSD for standard tests,
using two PCs. Testing with more PCs, with non-standard workloads,
and/or on a regular basis requires writing scripts and can get pretty
evolved (which let's us sell a pre-configured appliance that does
Polygraph test management :).

How-Tos for standard tests on FreeBSD are available at:
	http://www.measurement-factory.com/support.html

> Yes, on website kernel patches are avaliable for tunning, but for new
> releases of 4.x this isn't necessary, all could be configure with kernel
> options and sysctls (for 4.8):
>
> 	options		MAXFILES=16384
> 	options		HZ=1000
> 	options		NMBCLUSTERS=32678
>
> 	kern.ipc.somaxconn=1024
> 	net.inet.ip.portrange.last=40000
> 	net.inet.tcp.delayed_ack=0
> 	net.inet.tcp.msl=3000

One of our kernel patches optimizes handling of 1000s of IP aliases
per FreeBSD box. The patch is required for older 4.x kernels to
perform at decent levels. IIRC, the patch does not work for recent
kernels, probably because of the SYN cache changes. I do not know
whether any alias-related optimizations are still needed for recent
kernels though. Perhaps the SYN cache solves the original scalability
problem.

> Rest is quite simple/well documented. Tests in theory could be run
> on one machine, so... And some nice looking results generated by
> web-polygraph:
>
> 	Without any proxy:
> 	http://garage.freebsd.pl/pm3-15-11-2k2
> 	With squid:
> 	http://garage.freebsd.pl/pm3-05-11-2k2
> 	http://garage.freebsd.pl/pm3-06-11-2k2
> 	With external proxy:
> 	http://garage.freebsd.pl/pm3-29-01-2k3

Please note that a couple of the results I looked at are invalid from
PolyMix workload rules/design point of view. The first thing to check
is that you have huge numbers of request in waiting queue, compared to
active transactions (shown on the same "xact_lvl" graph). Most likely,
you overloaded the device under test, and most request ended up in
queues instead of on the wire.

I may be missing something though -- I am just looking at your
results without much knowledge of their history/purpose... See last
cache-off results for valid examples:
	http://www.measurement-factory.com/results/


If you have any Polygraph-specific questions, I would be happy to
answer them, especially if it can help FreeBSD folks in any way.

Good luck,

Alex.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.53.0302162130370.46493>