RE: Fw: Welcome to the InterNAT...

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



At 10:14 PM 3/26/2003 +0100, Jeroen Massar wrote:
Seeing that route filtering only gets done automaticaly for
the last couple of years and the fact that that is only a
route + ASN mapping I don't see why all of a sudden there
will be some magical solution for renumbering complete networks.

Really? I started to design such a procedure, and it's not terrifically hard if you set your sights correctly. The key concepts are:


- IPv6 makes it easy to have multiple prefixes and addresses per interface
- anything you have advertised in DNS etc has a lifetime, which you need to
honor
- the existing so-called renumber scheme mostly says how to make a new address
come into existence


So if you want to renumber from prefix A to prefix B, you have to:
-1- start with all routers and hosts using prefix A
-2- reconfigure all router interfaces to support both prefixes A and B,
and preferring prefix A
-3- as a side effect, you will also be starting to route for prefixes A and B.
-4- as a side effect, hosts will develop addresses in both prefixes
-5- Update DNS etc databases with prefix B addresses (presumably a
side-effect of step 4)
-6- at this point, there will be no new sessions opening using prefix A
as destination addresses. Prefix A will be in use in source addresses.
-7- reconfigure all router interfaces to support both prefixes A and B,
and preferring prefix B
-8- wait for all advertisements of prefix A addresses (DNS etc) to expire
-9- at this point, there will be no new sessions opening using prefix A
as destination addresses. Prefix A will be in use in source addresses
in sessions that started before step 6. Continue waiting until they
are no more (may be already complete, may take a month, YMMV)
10- when said sessions have all expired, prefix A is routed and usable but
is no longer in active use.
10- reconfigure all routers to support only prefix B
11- as a side effect, routing and host use of prefix A will cease


The hard parts are "reconfigure all routers..." and determining the duration of the delay in step 9. The rest actually works reasonably well today, I should think.



[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Fedora Users]