Re: [Bloat] Bufferbloat in local networks...

[Dave Taht <dave.taht@gmail.com>]

On Mon, Jul 22, 2013 at 3:34 PM, Robinson, Andrew 1. (NSN -
US/Mountain View) <andrew.1.robinson@nsn.com> wrote:
> Hi,
>
> Are there any tools to measure bufferbloat in local networks?

The most popular tool for local networks at this point is
netperf-wrapper's partial implementation of the rrul test suite.

https://github.com/tohojo/netperf-wrapper

It only implements about 1/3 of the specification at the moment, but
it, when also run with tools like mtr, are helpful for identifying
sources of bufferbloat on a local network, if you find any. It is
dependent on a recent version of netperf (2.6 or later) being
installed on client and server.

One misleading thing about the current rrul benchmarks is that it uses
ping rather than one way delay measurements, and that the udp traffic
generated stops after one loss. What I'd wanted was to be measuring
against an isochronous stream rather than something that scaled
geometrically with RTT.

On that front I have been evaluating owamp's methods and code (in the
hope of coming up with something simpler), but so far that's the
winner for creating and measuring one way delay and simulating voip
and gaming traffic.

http://www.internet2.edu/performance/owamp/

Most of my talks use the rrul suite in some way or another, although I
think I've been making a mistake in almost always presenting the 8
stream benchmark (I use the single stream stuff a lot)...


>
> Thanks,
> Andrew
>
> _______________________________________________
> Bloat mailing list
> Bloat@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
>



-- 
Dave Täht

Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.html