Re: "Internet Task Force Recognizes IPX" (sic)

[don provan <sgiblab!novell.com!troi![e-mail]>]

In article <[e-mail]> [e-mail] (William Manning) writes:
>Multiprotocol maybe.  Well connected NOT.  The big problem is the
>assignment and managment of XNS network numbers.

IPX is *based* on XNS and shares its headers with XNS, but it is an
entirely different protocol, so the management of XNS network numbers
is irrelevant.  This doesn't mean IPX network numbers don't have the
same problem (although Novell's looking into that), i just wanted to
clear up any possible misunderstanding.  (I'd hate to have everyone
beat up on Xerox until they turned over their XNS network numbers to a
central authority, only to discover this accomplishes nothing in the
IPX world.)

>I can't rout IPX over
>ANY of the backbones that I am aware of... can you? 

You can tunnel anywhere you'd like, as long as you can agree
on address assignment with all the other tunnel participants.

>When Xerox/IEEE [sic]
>hand over the XNS [sic] network number database to IANA, I'll start to believe
>in IPX as a real part of the the transit Internet.  For now, it is at best
>a local, leaf system protocol.

I wish it were confined to the leaves, but with the number of
questions i hear about Novell's products which tunnel IPX packets over
IP networks, i'm afraid you may be mistaken.  *Global* agreement of
network addresses is not required as long as there is agreement among
all connected parties of a given catenet.  This lack of central
control certainly has its limitations, but on the other hand, IPX
network numbers are not going to run out in the next six months, nor
even the next six years.
						don provan
						[e-mail]