[p2p-hackers] Real-world UPnP stats

Saikat Guha saikat at cs.cornell.edu
Mon Jun 5 14:23:31 UTC 2006

On Mon, 2006-04-24 at 11:56 -0700, Alex Pankratov wrote:
> We've recently added UPnP support to our client software and
> now I got some server-side stats and they are most interesting.
> Check this out -
> Roughly a half of all clients that reported success talking to
> their 'routers' and establishing TCP/UDP port mappings were
> still inaccessible from an outside via their mapped ports.

Interesting. Some reasons for this I can think of are as follows:

- If I recall, UPnP on a NAT that is behind yet another NAT doesn't work
well (since the UPnP Internet Gateway Device on most NATs 
typically doesn't act as a client for the outer NAT's UPnP IGD). This
comes into play when some connects a wireless router to their existing
wired-only NAT or something.

- UPnP doesn't have nice crash semantics. If an application registers a
mapping, crashes, is restarted, and re-registers the same mapping, some
NATs ignore the new registration while other NATs silently reap the old
mapping. The first case would result in broken behavior.

- On the flip side, if two different applications register the same
mapping, the latter mapping may silently override the former mapping or
alternatively, the latter may get silently ignored. Either way, one of
the applications suffers.

> Anyone care to comment or compare this to their own numbers ?

While I don't have numbers, I suspect UPnP success rate (for NATs that
support UPnP) would depend on the application as well as the topology
common for the target user population.

As a side note, if you want to test UPnP functionality, target Japanese
users -- based on anecdotal evidence, UPnP seems far more popular in the
far east than anywhere else.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: This is a digitally signed message part
Url : http://zgp.org/pipermail/p2p-hackers/attachments/20060605/04ef8204/attachment.pgp

More information about the P2p-hackers mailing list