Re: RFC 4666 M3UA - Reg Request Message

Lincoln,

All comments on any of the extension drafts is always welcomed.  The
LOADSEL and LOADGRP drafts are certainly not cast in concrete and can be
augmented or changed to meet whatever requirements the WG sees fit.  I
urge the WG to review the LOADSEL/LOADGRP drafts and consider whether
they can act as a basis for moving forward.

If we proceed with an extension draft approach, I urge the WG to
consider the procedures described in the ASPEXT draft which describes a
mechanism that allows an ASP to identify in the ASP Up (and SGP in the
ASP Up Ack) message which extensions they support.

 http://www.ietf.org/internet-drafts/draft-bidulock-sigtran-aspext-04.txt

However, I do agree that the requirements really need to be established.

<rant>
For example, I really believe that any perceived need for CIC ranges is
really just bad or lazy MGC engineering: hoping that the SG will solve
all the state sharing and redundancy issues for the MGC instead of
dealing with those things inside the MGC architecture.  That is, once
messages are delivered to a pair of redundant ASPs at the MGC, those
ASPs can distribute the messages across MGC blades, processors, nodes,
as the MGC designer sees fit without having to involve the SG at all.  A
couple of blades acting as ASPs with a hardware cost less than my
daughter's MP3 player could easily distribute the traffic for most of
the West Coast of the Americas.  Also, I don't believe that it was ever
anyone's intentions to distribute a single MGC along the 49th parallel,
CIC range by CIC range.  All the nodes serving the CIC ranges are at the
same location, likely in the same bay or shelf.  It is foolish in such
an arrangement to expect the loosely coupled arrangement between SG and
ASPs to be able to match the fail-over capabilities of a system that can
be tightly-coupled blade-to-blade, shelf-to-shelf and crate-to-crate.
News Flash: SIGTRAN will not make the redundant distributed system
design of your MGC unnecessary, and will not turn dumb ASPs into a
carrier grade system.
</rant>

--brian

Haresign Lincoln wrote:                      (Tue, 07 Nov 2006 14:54:18)
> Brian,
> 
> You're right, that will work.  
> 
> I'm in agreement with you that adding CIC to the list of RK won't work
> and we either need to further modify M3UA, or try to get your
> LOADSEL/LOADGRP drafts in alignment with all the requirements that might
> exist for all users (if they are not already).
> 
> Regards,
> Lincoln 
> 

-- 
Brian F. G. Bidulock
[email protected]
http://www.openss7.org/