prefix lengths [was Re: stla registry db issue]

[Bill Manning]

% Here's a somewhat belated response on this topic, and on the previous
% thread.
% 
% In this discussion it seems to me that the critical question is, what
% defines an ISP?

	And to answer this question it would be prudent to revisit 
	the discussions on this very point in the CIDRd and PIARA
	wg minutes from 1994-1997.  It turns out that the only
	significant difference between an "ISP" and and "RIR" is
	whether they announce via a routing protocol some/all of their
	delegated prefixes to peers.  

% themselves an ISP (and we don't care) their assignment will be "provider
% aggregatable" and not entitled to be announced globally. Rather, like any
% other downstream customer, their prefixes will be aggregated by the upstream
% provider within its own announcement.

	the kicker is, once a delegation is made, what enforcement
	methods are in place? And given the fluid nature of the 
	technology, fixing "entitlements" to specific delegations is
	downright silly.

% is in fact the proposal of the RIRs that the prefixes announced by any
% organisation will correspond only to their allocation (anywhere from /16 to
% /35), and not to their reservation (which would be either /16 or /29). The
% justification behind this (as discussed with the IAB in Minneapolis last
% year) is that in a future scenario where many TLA and subTLA blocks were
% released and allocated, and where routing table expansion again becomes a
% concern, it is necessary for ISPs to have an objective means available for
% limiting the size of their routing tables, and the best such means available
% is to filter on prefix length.

	there is zero enforcement capability for this view and attempts
	to do so, globally,  by the RIRs will get the RIRs into some 
	interesting discussions on restraint of trade, at least in the US.
	ISPs (or those announcing some or all of their delegated prefixen)
	can use these objective means, ON A CASE BY CASE BASIS, to determin
	what/if they will filter.

% The alternative is a huge routing table populated with /16 and /29 prefixes
% only, where no objective means for route filtering is available, and where
% bilateral routing agreements would emerge as the only way for an ISP to
% control its tables.

	The picture is not that bleak. While there is a push for migration
	to strict bilateral agreements, there is an opening to use 
	contracted, neutral route servers to construct routing views 
	based on ISP whim/policy for given points in the topology.
	Several folks use this service today and I expect it to become
	more widely used... unless the market collapses into a small
	handful of providers.  Quite frankly, I prefer the dense mesh
	of 100s of thousands of providers with peering relationships with 
	hundreds of peers instead of hundreds of providers with tens of peers.

	i.e. address delegation policy should not be based on a 
	presumption of potential peering, at least for IPv6.

	Sorry for the rant. 

% Regards,
% 
% Paul Wilson
% APNIC.
% 
% ______________________________________________________________________
% Paul Wilson, Director-General, APNIC               <pwilson@apnic.net>
% http://www.apnic.net                          ph/fx +61 7 3367 0490/82
% ----------------------------------------------------------------------
% See you at APRICOT 2000!  28 Feb - 2 Mar  http://www.apricot2000.ne.kr
% ----------------------------------------------------------------------
% 
% 
% 
% > -----Original Message-----
% > From: owner-6bone@ISI.EDU [mailto:owner-6bone@ISI.EDU]On Behalf Of Brian
% > E Carpenter
% > Sent: Wednesday, 29 December 1999 3:51
% > To: 6bone@ISI.EDU
% > Subject: prefix lengths [was Re: stla registry db issue]
% >
% >
% > Folks,
% >
% > 2**64 is a big number. It's the square of 2**32 if nobody
% > noticed. The majority of
% > BigCos will be able to understand this and use no more than an
% > SLA. If there are a few
% > idiot CIOs who insist on more for no good reason, it isn't the
% > end of the world.
% >
% > I am very relaxed about /29s being reserved at this stage of the
% > life of IPv6,
% > because 2**29 is also a big number. I'm not recommending any
% > change in the RIR
% > guideline of only allocating /35s; all I'm doing is saying that
% > we must stick
% > to the rule of not splitting /29s between ISPs.
% >
% > If BigCo is 20-homed, and doesn't want to deal with 20 prefixes,
% > then I can certainly
% > see a case for them leasing a prefix that can be in the
% > default-free table. But this
% > really will be the exception case. What we must do is ensure that
% > a 2-homed site can
% > easily deal with 2 prefixes. BTW, how many 6bone sites are
% > multihomed today?
% >
% >    Brian
% >
% > "Perry E. Metzger" wrote:
% > >
% > > "Michael H. Lambert" <lambert@psc.edu> writes:
% > > > On Thu, 23 Dec 1999, Bill Manning wrote:
% > > > >
% > > > >     Er, that is pretty much exactly the point I was trying to make.
% > > > >     If Brian is right and that group is successful in restricting
% > > > >     announcements to /29's, how much space is wasted for the sixty
% > > > >     nodes that form the cluster "www.bigco.com" that has connections
% > > > >     to 20 major ISPs?
% > > >
% > > > But is "bigco.com" a transit IPv6 provider?  My understanding
% > is that if
% > > > it isn't, it should never be allocated its own TLA.  It
% > should receive a
% > > > small block from each of its ISPs.  Or am I missing something?
% > >
% > > Anyone out there who thinks they can actually prevent GM or Yahoo or
% > > the like from getting their own routes announced should talk to an
% > > anti-trust lawyer.
% > >
% > > Perry
% >
% > --
% > - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
% > Brian E Carpenter (IAB Chair)
% > Program Director, Internet Standards & Technology, IBM
% > On assignment for IBM at http://www.iCAIR.org
% > Attend INET 2000: http://www.isoc.org/inet2000
% > Non-IBM email: brian@icair.org
% >
% >
% >
% 
% 


-- 
--bill