Re: M3UA AS override mode from SGP perspective

Chris Benson <[email protected]> Mon, 11 Oct 2010 11:02:09 -0700 (PDT)
Newsgroups gmane.ietf.sigtran
Message-ID <[email protected]>
Brian,

Thank you for your correction.  I wasn't trying to compare
two associations from two ASPs with two associations from
one ASP (only one SGP).  I was comparing two associations
between one ASP SGP pair, with just one association.

I am familar with M3UA implementations (and MTP3-user
applications) that use SLS values (along with other
fields) to expect ordered delivery within a single
data stream (usually by mapping it to one SCTP Stream).
I thus considered that this ordered delivery works when
only one association is available to reach the selected
SGP, and likely will not work when either of two different
associations might be used by M3UA to connect to reach the
same remote SGP.

The above usage scenario suggests only one association
per ASP SGP pair, which agrees with your observation and
recollection of the intent and meaning of the RFCs to
restrict to one association per pair of M3UA peer processes.

Thanks again, from Chris Benson.

On Sat, 9 Oct 2010, Brian F. G. Bidulock wrote:

>>  Date: Sat, 9 Oct 2010 02:48:04 -0600
>>  From: Brian F. G. Bidulock <[email protected]>
>>  To: Chris Benson <[email protected]>
>>  Cc: Andrew Booth <[email protected]>,  <[email protected]>,  <[email protected]>,
>>      <[email protected]>
>>  Subject: Re: [Sigtran] M3UA AS override mode from SGP perspective
>>  
>>  Chris,
>>  
>>  Chris Benson wrote:                          (Fri, 08 Oct 2010 13:03:29)
>>  > 
>>  > I am alsmost sure that the answer to this is yes, because
>>  > the assumed ordered capabilities of the underlying transport 
>>  > mechanism will no longer be true.  Messages in one association
>>  > can of course overtake messages in another association.
>>  
>>  It makes no difference whether two associations terminate on the
>>  same ASPs or different ASPs with regard to ordering of messages.
>>  
>>  > If that last viewpoint of mine is true (things will break),
>>  > then "no redundant associations" should be treated as a
>>  > requirement.
>>  
>>  If your conlusion where valid, M3UA could not use multiple data
>>  streams in an SCTP association.
>>  
>>  --brian
>>  
>>  -- 
>>  Brian F. G. Bidulock
>>  [email protected]
>>  http://www.openss7.org/
>>