Re: P2P Filesharing over UDP

David Barrett <[email protected]> Sun, 19 Jun 2005 16:42:35 -0700
Newsgroups gmane.ietf.nat.behave,gmane.ietf.midcom
Message-ID <[email protected]>
Pyda Srisuresh wrote:
> --- Saikat Guha <[email protected]> wrote:
>>Subsequent Rodi clients perform a port scan on a remote address as
>>before. If the address is a NAT'ed Rodi client (that has already
>>performed step 1 above), then one of the port-scan packets make it
>>through to the other end. This is assuming the NAT is Full Cone for UDP.

Just curious, how "broad" is this port scan?  Does it do a total 
portscan (ie, of the complete range of possible UDP ports), or pick a 
subset that it somehow knows will work?

It'd seem that firewalls and routers would quickly identify this type of 
traffic as malicious, and then block the scan from completing.  Does 
Rodi do anything to somehow evade being flagged and blocked as such 
(maybe random delays or non-sequential scanning)?

Regardless, thanks for the great Rodi overview.  I agree this is further 
evidence that full-cone is the most P2P friendly, address-restricted 
less so, and symmetric even less than that.  The closer we can get to a 
total full-cone world (or even just a non-symmetric world), the better.

-david