Opened 5 years ago

Closed 2 years ago

#12892 closed task (fixed)

Run KIST on some fast live relays, measure congestion

Reported by: robgjansen Owned by:
Priority: Medium Milestone: Tor: unspecified
Component: Core Tor/Tor Version:
Severity: Normal Keywords: tor-relay kist simulation analysis research-program
Cc: Actual Points:
Parent ID: #12541 Points: large
Reviewer: Sponsor:

Description

We should run vanilla Tor and KIST Tor on relays of similar capacities, and compare how KIST affects congestion and performance. Moritz agreed to provide some fast relays/exits for this, and Micah Sherr at Georgetown has a pair of 40 megabit/s non-exit relays.

We'll need libkqtime to measure kernel input/output queue congestion in real time:
https://github.com/robgjansen/libkqtime

Child Tickets

Attachments (6)

qtimes.shadow.results.pdf (252.3 KB) - added by pastly 2 years ago.
live.clientperf.shadow.results.pdf (172.9 KB) - added by pastly 2 years ago.
4week.live.clientperf.shadow.results.pdf (154.3 KB) - added by pastly 2 years ago.
4week.qtimes.shadow.results.pdf (252.9 KB) - added by pastly 2 years ago.
kist-032-72hr.qtimes.results.pdf (279.5 KB) - added by pastly 2 years ago.
kist-032-72hr.clientperf.results.pdf (185.4 KB) - added by pastly 2 years ago.

Download all attachments as: .zip

Change History (30)

comment:1 Changed 5 years ago by robgjansen

Summary: Run KIST on some live exits, measure congestionRun KIST on some fast live relays, measure congestion

comment:2 Changed 5 years ago by nickm

Component: - Select a componentTor

comment:3 Changed 5 years ago by nickm

Milestone: Tor: 0.2.6.x-final

comment:4 Changed 5 years ago by nickm

Milestone: Tor: 0.2.6.x-finalTor: 0.2.7.x-final

I'm tentatively bumping KIST stuff to 0.2.7.x, since I think it won't be done this month. Please let me know if I'm wrong

comment:5 Changed 5 years ago by nickm

I'm tentatively bumping KIST stuff to 0.2.7.x, since I think it won't be done this month. Please let me know if I'm wrong

comment:6 Changed 5 years ago by nickm

Owner: robgjansen deleted
Status: newassigned

comment:7 Changed 5 years ago by nickm

Keywords: 027-triaged-1-out added

Marking triaged-out items from first round of 0.2.7 triage.

comment:8 Changed 5 years ago by nickm

Milestone: Tor: 0.2.7.x-finalTor: 0.2.???

Make all non-needs_review, non-needs_revision, 027-triaged-1-out items belong to 0.2.???

comment:9 Changed 4 years ago by nickm

Milestone: Tor: 0.2.???Tor: 0.2.8.x-final

comment:10 Changed 4 years ago by nickm

Points: large

comment:11 Changed 4 years ago by robgjansen

Severity: Normal

This is blocked until #17598 is completed, and the code from https://trac.torproject.org/projects/tor/ticket/12890#comment:30 is merged into nickm's branch.

Last edited 4 years ago by robgjansen (previous) (diff)

comment:12 Changed 4 years ago by nickm

Milestone: Tor: 0.2.8.x-finalTor: 0.2.???

It is impossible that we will fix all 252 currently open 028 tickets before 028 releases. Time to move some out. This is my first pass through the "assigned" tickets with no owner, looking for things to move to ???.

If somebody thinks they can get these done before the 0.2.8 timeout, please assign it to yourself and move it back?

comment:13 Changed 3 years ago by teor

Milestone: Tor: 0.2.???Tor: 0.3.???

Milestone renamed

comment:14 Changed 3 years ago by nickm

Keywords: tor-03-unspecified-201612 added
Milestone: Tor: 0.3.???Tor: unspecified

Finally admitting that 0.3.??? was a euphemism for Tor: unspecified all along.

comment:15 Changed 3 years ago by nickm

Keywords: tor-03-unspecified-201612 removed

Remove an old triaging keyword.

comment:16 Changed 3 years ago by nickm

Keywords: 027-triaged-in added

comment:17 Changed 3 years ago by nickm

Keywords: 027-triaged-in removed

comment:18 Changed 3 years ago by nickm

Keywords: 027-triaged-1-out removed

comment:19 Changed 3 years ago by nickm

Status: assignednew

Change the status of all assigned/accepted Tor tickets with owner="" to "new".

comment:20 Changed 3 years ago by nickm

Keywords: tor-relay kist simulation analysis research-program added

comment:21 Changed 2 years ago by pastly

I will attach the graphs I shared at Wilmington after this comment.

The data was collected over 48 total hours: 1 day of KIST, 1 day of vanilla Tor ("AMAP" or As Much As Possible).

We've had issues reproducing the results with a 4 week total collection period and are trying to find time to look into it a little more. Frustratingly, it could be many things outside our control.

"shadow" in the file names is a misnomer.

The 1st time to last byte page in live.clientperf.shadow.results.pdf is for web downloads. The 2nd is for bulk.

Last edited 2 years ago by pastly (previous) (diff)

Changed 2 years ago by pastly

Attachment: qtimes.shadow.results.pdf added

Changed 2 years ago by pastly

Changed 2 years ago by pastly

Changed 2 years ago by pastly

comment:22 Changed 2 years ago by pastly

We've had issues reproducing the results with a 4 week total collection period and are trying to find time to look into it a little more. Frustratingly, it could be many things outside our control.

See the attachments prefixed with '4week'

comment:23 Changed 2 years ago by pastly

I rebased kist onto master recently. This new kist is what dgoutlet is reviewing and will be merged soon.

We wondered if the issue reproducing our results had to do with the extra_space addition in the kist per-socket limit we made. So for 24 hours each, we ran amap, kist-with (the change), and finally kist-without (the change).

As you can see in the graphs I'll attach momentarily, kist was better than amap with and without the change. So we recommend leaving it in the code.

I've just started a 2 week test (1 week amap, 1 week kist-with) that will hopefully return the same results.

Changed 2 years ago by pastly

Changed 2 years ago by pastly

comment:24 Changed 2 years ago by pastly

Resolution: fixed
Status: newclosed

I've just started a 2 week test (1 week amap, 1 week kist-with) that will hopefully return the same results.

It did.

Note: See TracTickets for help on using tickets.