On 8/7/2001 at 15:14:02 +0100, Randy Bush said:
Take a look at the mpls@uu.net archives for a rather exhaustive and exhausting discussion of this very subject, or provide facts as to what specifically doesn't scale.
to paraphrase vijay, if dealing with one rib is a major discussion in the operator and ietf communities, how many thousands of them do you think a prudent operator wants to deal with in 2547?
And on some agricultural list somewhere, they're discussing the proper way to raise a sheep. If there's all that discussion about how to raise one sheep, who wants to deal with a flock? Once you know how to deal with one big table, dealing with a bunch of tiny, simple tables is pretty easy, especially since, once you build them, they mainly take care of themselves. This really isn't that much to it, folks. If you're very brave and very careful, you can let your customers handle their routes. Otherwise, you can statically route to them. If you don't want to handle all those routes in your VPN networks, how many more plain IP customers are you going to handle? Configuring routing for an RFC 2547 MPLS VPN is not really much harder than configuring a plain IP customer. It can certainly be automated just as easily. Perhaps the issue here is, the folks arguing against the complexity are targeting their arguments on the people who really don't need to worry about it: the small network operators, who have a small support staff trying to run a BGP-enabled network. These guys might be MPLS VPN customers, but there's no need for them to be providers. To leech another person's metaphor here, the argument is that a 747 is too hard for your average pilot to fly. Fortunately, your average pilot doesn't have to fly it. -Dave