[tahoe-dev] I would like to know about the performance with different segment size and pipeline depth

Zooko O'Whielacronx zooko at zooko.com
Tue Jan 4 06:36:36 UTC 2011


Folks:

Last summer a few folks including Kyle Markley, Nathan Eisenberg,
François Deppierraz, and Terrell Russell did some benchmarking as part
of the development of Tahoe-LAFS v1.8.0 (which had a completely new
downloader implementation). Recently I was wondering if we could
replicate some of their results on other people's networks (or even on
their own networks again) and get more information about using
different segment size and different pipeline depth:

http://tahoe-lafs.org/pipermail/tahoe-dev/2010-August/005007.html

http://tahoe-lafs.org/pipermail/tahoe-dev/2010-August/004853.html

http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1264

http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1170

If you want to do this, be aware that there is a lot of variance in
the performance of a multi-way networking application like this. It is
hard to tell whether the performance results indicate some effect of
the code or the settings or if they just reflect unrelated network
traffic, the phase of the moon, etc. Maybe you could mitigate this by
running each measurement five times in a row and report all five of
the results of each measurement.

Regards,

Zooko



More information about the tahoe-dev mailing list