Re: FW: LS on Clarification of M3UA usage in 3GPP networks

wuhongjian,

Please see comments below...

wuhongjian wrote:                      (Thu, 28 Feb 2008 15:43:43)
> 
>    Hi, brian:
> 
>    First,thank  u  for  your patient in giving your explanation again and
>    again.
> 
>    In    scenario    1,   Provided   that   the   figure   changes   into
>    IPSP1--IPSTP--IPSP2,  when  IPSP2  becomes  inactive,  does  RFC  4666
>    specify  clearly  that IPSTP  has to send DUNA to IPSP1? I don't think
>    so. You  have told us, IPSP-IPSP is a point to point configuration, in
>    which    IPSTP    is    excluded.    But    IPSP1--IPSTP--IPSP2    and
>    IPSP1--IPSTP1--IPSTP2--IPSP2    are    the    most    needed   network
>    configurations  in  large  IP-based signalling network. We can't slide
>    over this issue.

There is no such thing as an IPSTP in the standard.  What you can
have according to the standard is an SG that is also an STP.  That
is what is shown.

However, as the SE IPSP is simply the following:

   ,-------.       ,-------.
   |       |       |       |
   | ASPF  |-      | ASPF  |
   |       | \     |       |
   '-------'  \    '-------'
               \       |    
   ,-------.    \  ,-------.
   |       |     \ |       |
   |  SGF  |      -|  SGF  |
   |       |       |       |
   '-------'       '-------'
    SE-IPSP         SE-IPSP

One could conceive of this vague IPSP-IPSTP-IPSTP-IPSP arrangement
as simply:

   ,-------.       ,-------.     ,-------.       ,-------.
   |       |       |       |     |       |       |       |
   | ASPF  |-      | ASPF  |     | ASPF  |      -| ASPF  |
   |       | \     |       |     |       |     / |       |
   '-------'  \    '-------'     '-------'    /  '-------'
               \                             /            
   ,-------.    \  ,-------.     ,-------.  /    ,-------.
   |       |     \ |       |     |       | /     |       |
   |  SGF  |      -|  SGF  |     |  SGF  |-      |  SGF  |
   |       |       |       |     |       |       |       |
   '-------'       :-------:     :-------:       '-------'
                   |       |     |       |                
                   | STPF  |-----|  STPF |                
                   |       |     |       |                
                   '-------'     '-------'                
    SE-IPSP         SE-IPSTP      SE-IPSTP        SE-IPSP

And, because the SGF (SG Function) is not needed on the SE-IPSPs and
the ASPF (ASP Function) is not needed on the SE-IPSTPs, this reduces
simply to:

   ,-------.                                     ,-------.
   |       |                                     |       |
   |  ASP  |-                                   -|  ASP  |
   |       | \                                 / |       |
   '-------'  \                               /  '-------'
               \                             /            
                \  ,-------.     ,-------.  /             
                 \ |       |M3PA?|       | /              
                  -|  SG   |- - -|   SG  |-               
                   |       |     |       |                
                   :-------:     :-------:                
                   |       |M2PA |       |                
                   |  STP  |-----|  STP  |                
                   |       |     |       |                
                   '-------'     '-------'                
      ASP             SG             SG             ASP   

Which is possible today according to the standard and is what I
diagrammed.  Now we had conceived of this SG-SG protocol based on
M3UA, once called M3PA that could connect SGs instead of using M2PA
between STPs, but it does not obviate the need for an STP function
and the existing standard also permits traditional SS7 links from
the STPs.


>    In  scenario  2,  according  to  your explanation, one MSCserver shall
>    populate     2     ASs:     AS1    with RK=DPC1+SI3(SCCP), AS2    with
>    RK=DPC1+SI13(BICC). And how many associations will MSCserver establish
>    with one  SG,  one  common  association or two separated associations?
>    IMHO,if  the  answer  is  the  former,  it  is  really complex; If the
>    latter,I think it is a kind of waste of resources.

You should know from the standard that only one SCTP association can
exist between an ASP and an SG.  It is not overly complex in this
case, rather it is necessary and sufficient: you need two AS
definitions because you need two state machines to track the
independent state of the two user parts.  If you do not need to
track the independent state of each user part, used a DPC RK,
resulting in only one AS state machine.

>    Furthermore,  after  reading the scenario 2 described in the Statement
>    from  3GPP,  I notice  that  it  is  quite  different from the one you
>    mentioned in your letters.

How so?  I don't see any differences significant to the discussion.

--brian

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