[metrics-team] Unusually slow OnionPerf measurements

[Rob Jansen]

> On Jun 6, 2017, at 10:35 AM, Karsten Loesing <karsten at torproject.org> wrote:
> 
> On 06.06.17 14:28, nusenu wrote:
>> Karsten Loesing:
>>> What changed on May 22?
>> 
>> it correlates with the increase of users:
>> https://twitter.com/nusenu_/status/870335474518425600
> 
> Aha.  Well, right, the increase could indeed be real.  Thanks!
> 
> All the best,
> Karsten


I think the more likely reason is that when we changed to OnionPerf, we added new OnionPerf client vantage points in new geographic locations. The new vantage points would experience unique performance that would affect the median and quartiles reported in the metrics graphs. To check this, we could look at the performance reported by a single vantage point that was running both the old Torperf and the new OnionPerf to see if the performance for that single vantage point changed after switching from TorPerf to OnionPerf.

(This has been on my mind for a while, finally remembered to respond.)

Best,
Rob
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.torproject.org/pipermail/metrics-team/attachments/20170720/5781468d/attachment.sig>