[metrics-team] TorPerf and OnionPerf

[Dennis Jackson]

Hi David,

Great to hear how the tracing work is progressing. I can suggest two
different solutions.

Firstly, I just gave a presentation on a new tool for benchmarking Tor
using real browser processes to the Mozilla-Tor Monthly meeting. It's still
in its early stages, but it is up and working. You could use it to drive a
Tor Browser or Firefox instance through Tor and use the web in a fashion
that approximates a normal user. I will send you a link to the video /
slides off-list in case you don't have it already.

Secondly, I think Onionperf uses TGen behind the scenes to generate
traffic. TGen should allow you generate data which looks like HTTP requests
for some level of 'looks like'. I'm not super familiar with the details,
but I think its pretty mature and there's some information (including a
HTTP TGen model) here: https://tmodel-ccs2018.github.io/ .

Best,
Dennis


On Tue, Aug 20, 2019 at 8:54 AM David Goulet <dgoulet at torproject.org> wrote:

> Greetings Metrics!
>
> If I'm not mistaken, TorPerf/OnionPerf only do bulk download measurements?
> In
> other words, only download a file of a specific size (50K, 1M...) and then
> timestamp various timings there.
>
> Does any of those tools have support for non bulk data that is transmitting
> burst of data at some random timings?
>
> Tor cell scheduling is, in theory, built to prioritize "quiet circuit" like
> let say an SSH session that sees some commands, not rsync type.
>
> Only bulk data download might not show us the full picture of latency on
> the
> network so thus my question.
>
> So that feature of "non bulk data" seems to me like a must have also?
>
> Some context: I'm currenlty investiguating the *inprocess* cell timings and
> queue priority within little-t tor which ultimately I would like to
> correlate
> with Torperf data or at least try to find some avenues to understand where
> congestion happens.
>
> Thanks!
> David
>
> --
> 22OYsO4jsRaRQWkUKLdtZ3Ol5nzIN+vauQE2JvQ8Q18=
> _______________________________________________
> metrics-team mailing list
> metrics-team at lists.torproject.org
> https://lists.torproject.org/cgi-bin/mailman/listinfo/metrics-team
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.torproject.org/pipermail/metrics-team/attachments/20190820/524e5758/attachment.html>