On Tue, 22 May 2001, Gert Doering, Netmaster wrote:
Hi,
(originally I did not really want to participate in this discussion, as much of it has already been said in the last LIR-WG meeting).
One thing got me thinking,though:
On Tue, May 22, 2001 at 02:16:32PM +0100, Carlos Friacas wrote:
/22 is a much more reasonable value by my view...
Be careful what you are asking for.
By lowering the "minimum" PA allocation for "non-ISPs" i wasnt thinking about getting more fragmentation... I was seeing that "minimum" <> "initial" too...
If we assume that minimum allocation size will go down to a /22, and further assume that one fourth of the full IPv4 address range will subsequently be handed out *and announced* as /22's, this means we will see ( 1/4 * 2^22 ) = 1048576 /22's announced in the global BGP table. That's over a million BGP routing table entries.
This will have a significant effect on BGP routing stability and also on the costs of global routing - you need a Gig of RAM in all the BGP routers (on distributed architectures, more than that). The CPU power required to handle a flap of a major line in a timely fashion (to keep BGP convergence times low) will be horrendous.
100% agree... i dont like to see it get much more bigger...
Also, it can be assumed that in this case, the global topology will become complex enough so that most of the time many of the smaller ASes won't be reachable anyway due to problems "on the way".
I think this is something I do NOT want to see.
So, what is my conclusion? I estimate that while IPv4 address exhaustion is going to be a problem (which IPv6 will solve), the routing topology will cause major problems *sooner* than IPv4 runs out, and we should do something against this. By this, I mean:
- strongly encourage people to renumber from historic PI space to PA space from their ISPs network block (and return the PI space to the RIRs, to be aggregated)
:-) the finest way would be turning it "mandatory"... i know this problem very well, as i've already done too much work recovering some 192.x.x.x still given by ARIN... :-) I still didnt finish this on my clients because some of them see it as a "RESOURCE", and some are trying to X*Y valid addrissing by only returning X address space... :-(
- stop handing out PI space
:-) well... could RIPE just talk to registered ISPs...? I think this is very, very hard thing...
- discourage "end users" from using multihoming with globally visible address space (there are other ways, like "get multiple uplinks to different POPs of the same ISP, and have them sign a SLA that will get you 99.9% reachability or money back").
this brings one question... is this way cheaper or not ? i've heard of some cases that a "end user" who is also a LIR got his address space "divided" (originated) between two ISPs and on the inside they somehow manage the situation... but this doesnt guarantee redundancy to their full address space, of course...
- discourage people from becoming LIR if that's only to get "portable" address space, with no intention of handing PA space out to customers.
no customers = no ISP ?
Yes, this might sound a bit harsh, but I'm *really* worried about routeability and reachability of anything in the next couple of years.
Yep... i see more than 104.000...
Now go and flame me... :-)
That's not the intention of it.
Gert Doering -- NetMaster -- SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299
./Carlos "Networking is fun!" ------------------- <cfriacas@fccn.pt>, CMF8-RIPE, Wide Area Network WorkGroup http://www.fccn.pt F.C.C.N. - Fundacao para a Computacao Cientifica Nacional fax: +351 218472167