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

Re: How to Run High Capacity Tor Relays

Thus spake Olaf Selke (olaf.selke@xxxxxxxxxxxx):

> Am 24.08.2010 17:27, schrieb Mike Perry:
> > It ends up taking about 2-3 days to hit an observed bandwidth of
> > 2Mbyte/sec per relay, but it can take well over a week or more (Moritz, do
> > you have a better number?) to reach 8-9Mbyte/relay. This is for an
> > Exit node.
> is the slow rampup really the case or rather an urban legend?

It was verified with 8 of my experimental nodes, and I believe Moritz
had quite a bit of difficulty getting to that level as well.

However, it would be great if we could get some hard data on how long
it has taken all of our fastest nodes to acquire data. If someone were
to write a script to process descriptor archives to check how long it
takes fast relays to ramp up, that would be really awesome. We'd
then be able to see exactly what effect Guard, Middle and Exit node
flags have on ramp up time:


> After about 24h the new exits blutmagie2, blutmagie3, and blutmagie4 I
> placed in service this May got the same amount of traffic like the old
> blutmagie. So rampup phase wasn't longer than one day.

Two factors may explain this:
1. Your exit policy is also more permissive than the ones we use. 

We use:

The default policy will likely cause a lot more traffic to be pushed
through your node due to P2P (ab)users.

2. If you reused keys, the bandwidth authorites would "remember" the
speed of your new nodes, and immediately vote them faster, which would
also helps a lot. It is only the very first appearance that takes a
really long time. 

Mike Perry
Mad Computer Scientist
fscked.org evil labs

Attachment: pgpCQviNkr4IS.pgp
Description: PGP signature