[tor-dev] Scaling bandwidth scanner results

[Matt Traudt]

I've made some good progress on a bare bones, doesn't-do-more-than-it-
has-to bandwidth scanner. It can generate output just like torflow[0].

We need to decide on how to scale results that come from different
measurement systems. The simple, don't-make-it-harder-than-it-has-to-be

Express each weight as a proportion of the total, and multiply by

some agreed total (e.g. for the current network it would have to be the
total of the consensus weight, but within some limited range to avoid
unbounded growth). 

So we need to:

1. Decide on a large total.

I suggest 50 million to start the conversation (bike shedding) based on
that being close to the current total consensus weight so relay
operators won't see a large (though inconsequential) change.

2. Have all the torflow operators switch to this new method.

Ouch. I wouldn't mind being told I'm wrong about this step being
necessary.

3. Find all the places that hint at consensus weight being directly
comparable to bandwidth (such as [3]) and change the wording.

Matt

[0]: https://paste.debian.net/1015409/
[1]: https://trac.torproject.org/projects/tor/wiki/org/meetings/2018Rom
e/Notes/BandwidthAuthorityRequirements#Scaling
[2]: https://trac.torproject.org/projects/tor/ticket/25459
[3]: https://gitweb.torproject.org/torspec.git/tree/dir-spec.txt#n2290

[Tom Ritter]

After #1 is decided, we can convert past bwauth data, can't we? If
it's helpful I can (at some point) compare your data against
historical (converted) data as I've been doing:
https://tomrittervg.github.io/bwauth-tools/

-tom

Post by Matt Traudt
I've made some good progress on a bare bones, doesn't-do-more-than-it-
has-to bandwidth scanner. It can generate output just like torflow[0].
We need to decide on how to scale results that come from different
measurement systems. The simple, don't-make-it-harder-than-it-has-to-be

Express each weight as a proportion of the total, and multiply by

some agreed total (e.g. for the current network it would have to be the
total of the consensus weight, but within some limited range to avoid
unbounded growth).
1. Decide on a large total.
I suggest 50 million to start the conversation (bike shedding) based on
that being close to the current total consensus weight so relay
operators won't see a large (though inconsequential) change.
2. Have all the torflow operators switch to this new method.
Ouch. I wouldn't mind being told I'm wrong about this step being
necessary.
3. Find all the places that hint at consensus weight being directly
comparable to bandwidth (such as [3]) and change the wording.
Matt
[0]: https://paste.debian.net/1015409/
[1]: https://trac.torproject.org/projects/tor/wiki/org/meetings/2018Rom
e/Notes/BandwidthAuthorityRequirements#Scaling
[2]: https://trac.torproject.org/projects/tor/ticket/25459
[3]: https://gitweb.torproject.org/torspec.git/tree/dir-spec.txt#n2290
_______________________________________________
tor-dev mailing list
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev

[teor]

Hi,

Matt, Juga, Nick and I discussed an alternative scaling scheme on
#tor-dev during the patch party.

Post by Matt Traudt
1. Decide on a large total.
I suggest 50 million to start the conversation (bike shedding) based on
that being close to the current total consensus weight so relay
operators won't see a large (though inconsequential) change.

Calculate the average consensus weight per relay.
And make it a consensus parameter.
Currently, it's about 7500.

bwauths scale results so that the total measured bandwidth is
approximately 7500 * (the number of relays they have measured).

This scales their results so they are similar to the current torflow
bwauths. (The current bwauths have total measured bandwidths
between 35 million and 70 million. Using this scaling scheme, a
bwauth that measured 7000 relays would have a total bandwidth
of 52.5 million.)

This scheme also allows a bwauth to scale and report partial
results: if it has measured half the network, it scales to 25 million.

Post by Matt Traudt
2. Have all the torflow operators switch to this new method.
Ouch. I wouldn't mind being told I'm wrong about this step being
necessary.

Under this alternative scheme, if the unscaled bwauths average
relay weight changes, we change the consensus parameter.

So we don't need to change torflow.

See this pad for our notes and calculations:
https://pad.riseup.net/p/n3mEulClf6ZV

T