[tor-talk] Reduced latency transport for TOR

[Mike Perry]

Thus spake Gregory Maxwell (gmaxwell at gmail.com):

> This work could be _very_ productive for future transport for TOR:
> 
> https://www.usenix.org/conference/nsdi12/minion-unordered-delivery-wire-compatible-tcp-and-tls
> 
> As opposed to a raw datagram transport it still gets through the
> firewalls and nats that TCP/TLS does and still looks like HTTPS to
> censorware.

FWIW, I spent the better part of today reviewing our UDP plans[1,2] and
pondering this design, but I think I've decided this is not terribly
useful to us right now, primarily because it would seem to require the
very same circuit crypto changes as we'd need for a full client-to-exit
datagram transport (to support reordering at the circuit crypto level).

1. https://research.torproject.org/techreports/datagram-comparison-2011-11-07.pdf
2. https://research.torproject.org/techreports/datagram-testing-plan-2012-03-16.pdf


-- 
Mike Perry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.torproject.org/pipermail/tor-talk/attachments/20121108/eb7a670a/attachment-0001.pgp>