Re: [v6ops] ULA draft revision #2 Regarding isolated networks

[Ted Lemon <[e-mail]>]

On May 27, 2014, at 5:24 AM, Philip Homburg <[e-mail]> wrote:
> I don't see why we would need to discuss the foreverness of ULA, when GUAs are not
> at all forever. 

You'd like to think that, and in some circumstances it makes sense.   If my homenet has a ULA, it has it so that local communications can proceed when I don't have an internet connection.   A renumbering event that switches to a new ULA shouldn't be a big deal.

The operational situation that's problematic is the large enterprise scenario, where you have two large enterprises with their own ULAs that merge.   If those ULAs happen to clash, you have to renumber at least one of them.   If they don't clash, you still have to deal with routing them (although I think the split-horizon complexity objection Mikael raised ought to be thought through carefully before being asserted as factual, because I think it can be addressed through routing and not naming).

I wish people wouldn't say "that is evil and must be burned" because it's clearly useful in some real scenarios where there is no better alternative, and hence I think documenting those scenarios and explaining how to do them right (IOW, not wind up with NAT) is a valuable service we can do for the community, which will use them regardless.