Re: SIP Addressing Limitations

William Allen Simpson <bill.simpson@um.cc.umich.edu> Fri, 21 May 1993 14:10 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa03968; 21 May 93 10:10 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa03964; 21 May 93 10:10 EDT
Received: from p.lanl.gov by CNRI.Reston.VA.US id aa03364; 21 May 93 10:10 EDT
Received: from noc-gw.lanl.gov by p.lanl.gov (5.65/1.14) id AA20763; Fri, 21 May 93 08:05:33 -0600
Received: by noc-gw.lanl.gov (4.1/SMI-4.1) id AA08022; Fri, 21 May 93 08:04:50 MDT
Return-Path: <bill.simpson@um.cc.umich.edu>
Received: from p.lanl.gov by noc-gw.lanl.gov (4.1/SMI-4.1) id AA08018; Fri, 21 May 93 08:04:48 MDT
Received: from vela.acs.oakland.edu by p.lanl.gov (5.65/1.14) id AA20725; Fri, 21 May 93 08:04:47 -0600
Received: from via.ws07.merit.edu by vela.acs.oakland.edu with SMTP id AA24963 (5.65c+/IDA-1.4.4); Fri, 21 May 1993 10:03:40 -0400
Date: Thu, 20 May 93 13:31:46 EDT
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: William Allen Simpson <bill.simpson@um.cc.umich.edu>
Message-Id: <1196.bill.simpson@um.cc.umich.edu>
To: "Paul Francis (formerly Paul Tsuchiya" <francis@thumper.bellcore.com>
MMDF-Warning: Parse error in original version of preceding line at CNRI.Reston.VA.US
Cc: pip@thumper.bellcore.com, sip@caldera.usc.edu, tuba@lanl.gov
Reply-To: bsimpson@morningstar.com
Subject: Re: SIP Addressing Limitations

> Given the following:
>
>    Luxembourg           0.39    0.5e  1.1  C001 0000 0111 0000 0... ....
>                                                      ---- ---- ---- ----
>    United Kingdom      57.8    61.0   0.3  C001 0001 0... ....
>    Ireland              3.5     5.0  -0.3  C001 0001 1000 0...
>    Guernsey&Jersey(UK)  0.16    0.2e  0.8  C001 0001 1000 1000 00.. ....
>    Isle of Man (UK)     0.064   <.1e  0.1  C001 0001 1000 1000 010. ....
     [skipped]                                                   ---- ----
     [Northern Ireland]                      C001 0001 1000 1001
>                                                       --- ---- ---- ----
     [2nd provider]                          C001 0001 1111 1111 1111 1110
     [1st provider]                          C001 0001 1111 1111 1111 1111
     France              56.9    58.6   0.4  C001 0010 0... ....
>
> UK has a 0 at the beginning of the second byte, and the other
> three have a 1 at the beginning of the second byte, thus I
> conclude that the UK is not clustered with the other three.
> But, this doesn't seem to follow common sense, so maybe I'm
> wrong.
>
The --- separate the clusters, and indicate bits that may be used for
provider assignment and future expansion.  To quote (p 3):

      This division provides the primary intersection between
      metropolitan and provider based allocation.  Approximately 1/4 to
      1/2 of the numbers in each cluster are reserved for provider based
      allocation and future expansion.  This is indicated by dashes (---)
      at the end of each cluster.  Large regional providers should be
      assigned from the high-numbered portion of each cluster.  New
      countries may be added to the low-numbered portion of each
      cluster.

There also was a section that you must have missed on the careful
attention paid to bit and octet alignment.  The cluster in this case is
C001 0001, which includes the U.K.  The next cluster begins with France.

The assignment is designed so that aggregation may occur during the
normal operation of the routing algorithms.  Viewed from France, the
cluster may not aggregate at all, since Guernsey and Jersey are closer
to France.  But viewed from Germany, it is extremely likely to
aggregate; from China, almost certain.

> Could you give me an example of how a provider would be
> assigned in the context of the above-mentioned cluster?
>
Providers which served more than one country in the cluster would be
assigned from the end of the --- area, as illustrated above.  If a new
country is defined (say Northern Ireland), it might be assigned as above
(assuming that is a good fit based on population).

Note that internal assignment is a concern only of the country.  Some
governments -- like the U.K. -- are notoriously recalcitrant about plans
designed in the U.S.

All this plan says is, at the country level, if you use metropolitan or
end-point based assignment, assign numbers for the metros from the low
end, and assign end-points by metro.

Providers which only serve a region within a country (say Scotland,
within the U.K.), would be assigned from the high end of the country
space, growing toward the others.

Bill.Simpson@um.cc.umich.edu