Date: Mon, 07 Sep 2009 15:49:15 +0200 From: Willem Jan Withagen <wjw@digiware.nl> To: Mike Tancsa <mike@sentex.net> Cc: net@freebsd.org Subject: Re: UDP output performance Message-ID: <4AA50F5B.1080600@digiware.nl> In-Reply-To: <200909071301.n87D1kqY062913@lava.sentex.ca> References: <4AA14018.3010102@digiware.nl> <5bc218350909041002x670460c8nf202a714182d1bf6@mail.gmail.com> <4AA4F11C.4060200@digiware.nl> <200909071301.n87D1kqY062913@lava.sentex.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
Mike Tancsa wrote:
> At 07:40 AM 9/7/2009, Willem Jan Withagen wrote:
>
>> Well this turns out to be a pilot error, in that I created such a
>> complex bandwidth evaluation that on buffer full the packet got tossed
>> in the application.
>> :(
>>
>> Just stripping that out, and just do a
>>
>> try send
>> while(not send) {
>> usleep(1 packet-time);
>> try again;
>> }
>
> There are some nice simples tool in /usr/src/tools/tools/netrate that
> are great for generating arbitrary bandwidth via udp packets you might
> want to have a look at.
Thanx Mike, I'll have a look. Always better well copied, than badly self
invented. (free to an old dutch proverb). :)
Reason I'm home growing, is that these tests are for really specific
applications we're developing.
And the actual purpose is to find hardware combo's that will do UDP packets
a fast at we can get it out of the ports.
Preferably with a 10Gbit card at close to 10Gb/s.
But thus far I'm only getting the intel hardware on 1Gbit close to the line
max. But I've got several boards still to test.
As I side note:
If anybody had experience with 10Gb hardware, I would appreciate
recommandations for that as well.
One of the piexes of hardware I'm currently evaluating is:
Which does a nice job, with 2mbit udp in on em0 and 1Gbit udp out on port em6.
--WjW
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4AA50F5B.1080600>