[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-bugs] #25687 [Core Tor/Tor]: over-report of observed / self-measure bandwidth on fast hardware -- important to torflow / peerflow



#25687: over-report of observed / self-measure bandwidth  on fast hardware --
important to torflow / peerflow
-------------------------------------------------+-------------------------
 Reporter:  starlight                            |          Owner:  (none)
     Type:  defect                               |         Status:  new
 Priority:  Medium                               |      Milestone:  Tor:
                                                 |  unspecified
Component:  Core Tor/Tor                         |        Version:  Tor:
                                                 |  0.2.6.10
 Severity:  Normal                               |     Resolution:
 Keywords:  tor-bwauth, needs-research, needs-   |  Actual Points:
  proposal?                                      |
Parent ID:                                       |         Points:
 Reviewer:                                       |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by starlight):

 After writing the above I realized that perhaps the recommendation should
 be that three separate bias curves be constructed:

 a) curve to affect the degree to which positive offsets of a relay's
 measurement to the average measurement is applied to the effective self-
 measure

 b) curve to affect the degree to which negarive offsets of a relay's
 measurement to the average measurement is applied to the effective self-
 measure

 c) curve to bias the weight result once either (a) or (b) have been
 applied to effective self measure

 (a) and (b) are a fancier version of Torflow's Kp value, and (c) is
 intended to allow a designed rather than accidental biasing of faster over
 slower relays that results from the ratio offset approach.  Since a moving
 average should remove the current behavior where fast relays have very
 high multiples over the all-relay average measurement, the (c) curve
 allows thoughtful restoration of a similar effect and IMO the combination
 of a moving average and the (c) bias curve can be used to solve the
 problem of outlier prone scoring w/r/t to individual relays.

 Perhaps (a) and (b) are overkill, and will end up as 0.95, 1.00 or 1.05 *
 x ^0^ or the like, but since it's easy to code putting it in entails
 downside and it could be useful.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/25687#comment:19>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs