>>>>> "Randal" == Randal L Schwartz <[email protected]> writes:

Randal> [...] Suppose I own 4.x.x.x right now (that's 4.0.0.0/8), and
Randal> I'm really only using half.  So I decide to sell you 4.N.x.x
Randal> where N > 127.  Now, if you're not immediately adjacent to me
Randal> on the internets (and it's not possible to be adjacent on
Randal> multiple vectors, so the answer is *never*), every single ASN
Randal> router will have to record *two* routes where they formerly
Randal> had one: one toward me for 4.0.0.0/9, and one towards you at
Randal> 4.128.0.0/9.

Randal> If people are allowed to do that, the routers would quickly
Randal> collapse.  The RIRs won't allow that.

Randal> Thus, there's not going to be an IPv4 gold rush, and instead
Randal> IPv6 *must* be deployed before the first RIRs are dry.

I'm not an expert on backbone routing, but ... I don't grasp how a few
extra ipv4 routes are going to lead to armegedon but ipv6 routing
isn't.  Surely with the proliferation of address space, ipv6 is going
to be *more* expensive to route than ipv4.

My little 200MHz 32meg RAM Mips-based Netgear WGT634Us are maintaining
routing tables hundreds of routes long and it doesn't seem to be
causing them much strain.

http://en.wikipedia.org/wiki/File:BGP_Table_growth.svg

Are you really suggesting that even 256 (instead of 2) new routes is
going to make a difference?


-- 
Russell Senior, President
[email protected]
_______________________________________________
PLUG mailing list
[email protected]
http://lists.pdxlinux.org/mailman/listinfo/plug

Reply via email to