RE: Multiple SUA SGs, sending for SSP from SGs

Stanislav,

-----Original Message-----
From: [email protected] [mailto:[email protected]]On Behalf Of
Stanislav Ivanovich
Sent: Friday, January 13, 2006 2:53 PM
To: SIGTRAN
Subject: RE: [Sigtran] Multiple SUA SGs, sending for SSP from SGs


Tolga, Brian,

I see that both of you agree that in this case at ASP place we must have
MTP3 (or SCCP) logic if one wants to use this configuration scenario. I
think no one can question this!
[TOLGA]Just to prevent a possible confusion, not a fullblown MTP3 logic
though, just enough to select the "right" SG.

However the essential problem is which protocol we are to use? In my view
this is the essence of the dispute. If we are to allow that ASP (or IPSP)
can contain MTP or SCCP logic then the difference between the processes
disappears (e.g. ASP/IPSP/SGP all can do relay/gateway/host applications) as
we saw from the discussion we had last year (see thread "[Sigtran] AS
concepts and Relay in SUA and M3UA" starting from December 14th 2005). So
why then different protocols (ASP-SGP and IPSP-IPSP in SE and DE flavors)?

However, according to Tolga's idea SG-SG protocol does support this -> i.e.
MTP3/SCCP layers communicate without AS concepts in between!
[TOLGA]First of all I have to say that the credit for SG-SG does not belong
to me alone. There were other people contributing to it as well from very
early on, when the idea came up during initial IPSP discussions -I believe
more than 4 years ago :-)-, or afterwards to review the SG-SG drafts.

What is the redundancy scheme which ASP-SGP or IPSP-IPSP communication can
give you and SGP-SGP communication cannot give you if in this cases
ASP/IPSP'es only contain MTP3/SCCP layers (i.e. without application logic)?
[TOLGA]From redundancy point of view, probably there is no major
difference -or better said, I personally don't see something which couldn't
be added to SG-SG in this regard, if there is a need-. AFAICS, the main
advantage of ASP/SGP model is to allow routing on AS level, i.e. in sub-PC
granularity.  OTOH, if PC boundaries are crossed, I find SG-SG model more
natural and efficient because it is a true peer-to-peer model and does not
require hosting MTP3/SCCP logic on a remote node on behalf of AS.

OTOH, I see some problems of utilizing AS concepts in between two MTP3/SCCP
layers (regardless of how you call the hosting processes)....
[TOLGA]Yes, here I agree with you and actually if one considers that ASP/SGP
interface is mimicing MTP3/MTP3-User Part (SCCP/SCCP-User Part for SUA)
interface, this is not a suprise. That type of interface is more suitable to
distribute traffic for a single signaling entity -whether it be MTP3 or
SCCP- to multiple application logic processing nodes.

/ Stanislav



Tolga Asveren <[email protected]> wrote:
Brian,

[..snip..]
> > [TOLGA]Even for PC granularity RKs, it is SG which controls/creates the
> > unique view of the signaling enitity to the network, so one has to have
> > necessary MTP3 procedures to support that.
> >
> > BTW, I wouldn't call hosting MTP3 logic in SG as "a P! C in MTP3
> stack in SG",
> > it is just providing necessary procedures to have the unique view of the
> > signaling enitity and generating corresponding network messages
> -and it is
> > the multiple clones of the same unique view on different SGs,
> which are not
> > synchronized, which may create issues for multiple-SG scenarios-.
>
> STPs do this all day long.
[TOLGA]On the practical side of how things work, I agree with you that with
multiple-SG acting as STP scenario where RKs are in PC granularity, the
possibly different views of the hosted PC are less of a problem. In such a
case hosting necessary MTP3 logic for AS maps pretty much to route
management from functionality point of view.



_______________________________________________
Sigtran mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/sigtran





Yahoo! Photos
Got holiday prints? See all the ways to get quality prints in your hands
ASAP.