Re: Multiple SCTP associations

Sergey,

I am not agreeing with Tolga much lately...

It was never our intention to allow more than one SCTP association
between a given pair of processes to persist.

E.g. RFC 3332 1.3.2.4:

 ... In order to [sic] avoid redundant SCTP associations between two
 M3UA peers, one side (client) SHOULD be designated to establish the
 SCTP association, or M3UA configuration information established to
 detect redundant associations (e.g., via knowledge of the expected
 local and remote SCTP endpoint addresses).

This passage makes it clear that measures are to be taken to avoid
multiple associations.

Also, an ASP Identifier is required whenever multiple ASPs exist at
the same endpoint.  If the provided ASP Identifier is missing,
ERR("ASP Identifier Required") is returned to ASP Up.  If the ASP
Identifier is not unique, ERR("Invalid ASP Identifier") is returned
to ASP Up.  Therefore, multiple associations to the same ASP will not
allow ASP Up to proceed on all but one of them.

This is the same for both M3UA (RFC 3332 Section 3.8.1) and SUA
(RFC 3868 Section 3.9.12).

--brian

On Wed, 11 Jan 2006, Sergey Mikhailov wrote:

> Tolga, Ilie,
> thank you for your valuable replies on the subject!
> 
> Please let me know, if I misinterpreted your answers.
> 
> 1.
> As far as I understand it, an SCTP endpoint, to say it strictly, may have 
> only one SCTP association; and a signaling process may use multiple SCTP 
> endpoints (and hence multiple SCTP associations) for communication with 
> different signaling processes of one AS/SG.
> 
> 2.
> Multiple SCTP associations between the same two signaling processes, while 
> not prohibited, will require reinterpretation of SCTP-to-ULP events (to keep 
> communication status and ASP states consistent across multiple SCTP 
> associations), such as NETWORK STATUS CHANGE, COMMUNICATION UP/LOST/ERROR, 
> RESTART and SHUTDOWN COMPLETE events (RFC2960).
> 
> 
> Regards,
> Sergey Mikhailov.
> 
> 
> 
> ----- Original Message ----- 
> From: "Tolga Asveren" <[email protected]>
> To: <[email protected]>
> Sent: Tuesday, January 10, 2006 7:33 PM
> Subject: RE: [Sigtran] Multiple SCTP associations
> 
> 
> > Sergey,
> >
> > -----Original Message-----
> > From: [email protected] [mailto:[email protected]]On Behalf
> > Of
> > Sergey Mikhailov
> > Sent: Tuesday, January 10, 2006 11:39 AM
> > To: SIGTRAN
> > Subject: [Sigtran] Multiple SCTP associations
> >
> >
> > Hi,
> >
> > It seems that multiple SCTP associations established between the same two
> > signaling processes are not [explicitly] prohibited.
> >
> > 1.
> > Is it legal for a signaling process to establish more than one SCTP
> > association to the same AS/SG where these multiple associations are served
> > by different signaling processes?
> > [TOLGA]Yes, this is how host redundancy is achieved (I assume you talk
> > about
> > M3UA/SUA). There are multiple ASPs serving an AS, and multiple SGPs in a
> > SG,
> > and there is a SCTP association between each SGP/ASP pair -well, obviously
> > it is not mandatory-.
> >
> > 2.
> > Is it legal for a signaling process to establish more than one SCTP
> > association to the same signaling process at the remote side?
> >
> > (I feel that the 2nd case may be reasonable only when the remote side does
> > not support multi-homed SCTP endpoints.)
> > [TOLGA]Multiple SCTP assocaitions between two etities will have problems
> > with ASP state machine if you follow it strictly considering the state
> > transitions associated with events from SCTP. This is easy to overcome
> > with
> > a some reasonable reinterpretation. OTOH, multiple SCTP associations won't
> > be a replacement for SCTP multihoming support because SCTP multihoming
> > provided network/NIC redundancy without loss/missequencing of messages.
> >
> > Regards,
> > Sergey Mikhailov.
> >
> >
> >
> > _______________________________________________
> > Sigtran mailing list
> > [email protected]
> > https://www1.ietf.org/mailman/listinfo/sigtran
> >
> 
> 
> _______________________________________________
> Sigtran mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/sigtran

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
[email protected]    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦