Re: Why One Standard IGP? (was "Intra-AS Routing Standards")

["Philip Almquist" <almquist at jessica.Stanford.EDU>]

Zaw-Sing,
	I probably should have resisted answering this, but my defenses seem
to be down and this is an issue I feel very strongly about...

> As long as they conform to the same standard, gateways presumably should
> interoperate.  If we have two [RECOMMENDED IGPs], just for the sake of
> argument, RECOMMENDED IGPs, a vendor may choose one or the other or both
> to implement depending on their market strategy.

	I chair the technical oversight committee for a regional network.  In
this regional, we use routers from two different vendors.  Both make fine
products.  Each supports a sophisticated IGP which some might propose as a
candidate for an Internet RECOMMENDED IGP.  Unfortunately, because of
differing market strategies, the sophisticated IGPs chosen by the two vendors
are not the same.  To make the routers interoperate, we have to use RIP, a
distinctly unsophisticated IGP which constrains our topology and connectivity
in severe and unfortunate ways (mostly because it can't tag routes by source
AS).

	I also design the networks at the INTEROP conferences.  In the most
recent one, we used routers from three different vendors, not counting the
ones that our exhibitors connected to the network.  Although some of them
supported sophisticated IGPs, the only IGP supported by all of them was RIP.
Unfortunately, RIP is so unsuitable in that sort of environment (because of
its lack of authentication) that we had to largely resort to static routing.
Perhaps the only thing less fun than using RIP is not being able to...

	Choosing a standard which allows vendors to choose from a list of
standard IGPs will only perpetuate situations like the above.  Routers still
won't interoperate unless their vendors have made compatible choices.  That
doesn't sound much different from not having a standard.

	WE NEED AT LEAST ONE GOOD MANDATORY IGP!  WE NEEDED IT LONG AGO!  THE
IAB SHOULD SERIOUSLY CONSIDER CHOOSING ONE NEXT WEEK!

	Whew!  I feel better now...

						Philip

disclaimer: The above comments do not necessarily represent the opinions of
the IETF Router Requirements Working Group or the sponsors of the INTEROP
conferences.  They probably do, however, represent the opinions of BARRNet.