On Jun 29, 2004, at 1:44 PM, Richard Welty wrote:
On Tue, 29 Jun 2004 13:32:30 -0400 (EDT) Jon Lewis <jlewis@lewis.org> wrote:
So, how do your filters tell the difference between these broken out NAC routes through a new provider and "multihomed customer routes with the primary provider's connection down"?
i've played this game from the multi-homed customer side before. you get your second provider to route the smaller space, and you expect the small announcements to be dropped by some ISPs and depend on the aggregate from your first provider to cover your bases there.
it only works as long as the first provider continues to provide transit.
It works as long as the first provider: 1) Continues to announce the aggregate, which NAC obviously will, and 2) Accepts deaggregates of his own space from peers, which the TRO requires NAC to do. (Not specifically, but if NAC filters this block, the judge almost certainly will find them in contempt.) If it is Pegasus and they have a /16, the point is moot. If it is some guy with a /24 out of non-swamp space, NAC will be providing transit for them. For instance, traffic from, say, Verio will be routed to the aggregate NAC announces, and NAC will have to pass it off to the new transit provider since Verio will not see the /24. This obviously has a cost to NAC, and it could be a high cost if this traffic goes over NAC transit in any real volume. IANAL, but seems like a Very Good Reason to not make the "T"RO permanent. -- TTFN, patrick