>> Aren't Microsoft already "standardizing" this with their Universal Plug and >> Play (UPnP) architecture? > > That's just midcom, which the IETF is standardizing. We > started before they did but Microsoft got there first and > worst (there's even midcom language in their documents). So > that's something that was started in the IETF and > expropriated. Note that it still handles incoming > connections pretty badly.
The UPNP "Internet Gateway Device" working group and the MIDCOM WG effort started at about the same time, but the UPNP specification was published about 2 years ago, while MIDCOM is not quite finished yet. Looks like an interesting case study for the "problem statement" working group... I don't quite get the point on the handling of incoming connections -- basically, the UPNP/IGD specification has functions for reserving port mappings, that allow a host behind a NAT to listen on an external port number. The host can definitely receive incoming connections. The UPNP/IGD approach has limitations: it is designed for a "home network" rather than an enterprise network, and it will not work if there are two "stacked NAT". However, it does solve a large fraction of the problem. By the way, while Microsoft certainly contributed to the fundation of the UPNP Forum, the UPNP/IGD is not exactly a Microsoft specification. -- Christian Huitema
