[p2p-hackers] UDP Hole Punching through Symmetric NATs

[Saikat Guha]

On Wed, 2005-06-15 at 03:09 -0700, David Barrett wrote:
> So hole punching through symmetric NATs is difficult, and Bryan suggests 
> not worth the effort.  I'm curious what you think, however.  Have you 
> tried it and found it useful in the real world?

We performed port prediction and hole punching tests, albeit for TCP.
The compiled results are available at:
http://nutss.gforge.cis.cornell.edu/pub/draft-imc-stunt.pdf

Detailed results of the port prediction can be found at:
https://www.guha.cc/saikat/stunt-results.php?sort=10,5,6

While this is for TCP, we found earlier that more NATs support hole-
punching and port prediction for UDP than for TCP; so interpret the
results as a lower bound or a qualitative indication of the trend.

With 82% "cone" NATs, as per Bryan's paper, the UDP P2P success rate is
slightly less since both sides need to successfully punch holes. FWIW,
we measured the effect of port prediction for TCP. With 75% "TCP cone"
NATs the P2P success rate without port prediction is roughly 50%; if we
add in port prediction for the symmetric NATs, the success rate
increases by 35%.

Cheers.
-- 
Saikat
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://zgp.org/pipermail/p2p-hackers/attachments/20050615/4a378a37/attachment.pgp