[tor-talk] IDEA: Compress traffic at exit

grarpamp grarpamp at gmail.com
Sun Jan 22 09:03:28 UTC 2012


>>    This is about compressing traffic at the exit, where it's passed
>> from outside the network through other relays to the client where it
>>    gets decompressed.
>
> And what decompresses the traffic at the destination site?

I think this is a question is of compressing traffic internally between
the exit or HS and the client. It's pretty clear Tor should not be
sending 'Accept: gzip' or whatever headers out past the exit or HS,
aka: the destination site.

I doubt such internal compression would help, because the bulk of
data behind the bandwidth is uncompressible... multimedia. Back
in the 90's when text was king and crypto was rare, sure, it would
have been a win.

> Lack of bandwidth isn't so much the problem. Lack of tcp sockets and cpu
> power for onionskins at the relays is the problem. The higher bandwidth
> the relay, the worse these two problems become.

Is there a bandwidth rate sweet spot at where dropping in a new non-exit
relay will soak up enough users to put its fairly massive new complement
of CPU/TCP resources to the network to good use? In other words, I figure
any PC has far more of those resources than bandwidth. Unless it is
perhaps an exit, or anchored to an HS where it's swamped with
connections as a focal point.


More information about the tor-talk mailing list