RE: Lossless fail-over

Hi Lincoln,

I am just presenting the rationale behind a conscious decision by SIGTRAN
WG. It could be that some fixes may be necessary. As my current personal
opinion, I do not think something similar to correlation Ids is necessary at
all.

In the same expired draft I mentioned, certain solutions are suggested for
cases where SCTP associations go down for reasons except host failure, e.g.
adjusting SCTP parameters.

    Thanks,
    Tolga

> -----Original Message-----
> From: Haresign Lincoln [mailto:[email protected]]
> Sent: Monday, June 19, 2006 9:17 AM
> To: Tolga Asveren; [email protected]
> Subject: RE: [Sigtran] Lossless fail-over
>
>
> Tolga,
>
> It is possible that associations fail for other reasons than loss of
> peer.  We see associations failures in overloaded or poorly designed IP
> networks.  I don't think we can always count on properly designed
> networks.
>
> Regards,
> Lincoln
>
> -----Original Message-----
> From: Tolga Asveren [mailto:[email protected]]
> Sent: Monday, June 19, 2006 8:47 AM
> To: [email protected]
> Subject: RE: [Sigtran] Lossless fail-over
>
> Jan,
>
> I will try to summarize the rationale behind not having
> changeover/changeback type of functionality in M3UA/SUA:
> (there is an expired draft about M3UA deployment considerations, if you
> are interested I can send you the relevant sections from it.)
>
> a) Network failures are handled by SCTP + properly designed network.
> b) Because of a) loss of SCTP association means peer going down.
> c) When peer is down, changeover/changeback doesn't help -also not in
> conventional SS7-.
>
>    Thanks,
>    Tolga
>
>
> -----Original Message-----
> From: Kolomaznik Jan [mailto:[email protected]]
> Sent: Monday, June 19, 2006 8:44 AM
> To: [email protected]
> Subject: [Sigtran] Lossless fail-over
>
>
> Hello everybody!
> Is there any document specifying lossless fail-over or change-back
> procedures in Sigtran especially in SUA?
>
> I found out only expired draft-bidulock-sigtran-corid-03.txt. >From my
> point of view this draft solves what I exactly need - but it is only
> expired draft. How to solve it on inter-vendor level? Does anybody know
> why this draft was leaved?
>
> Thanks
> Jan Kolomaznik
>
>
> _______________________________________________
> Sigtran mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/sigtran
>
> _______________________________________________
> Sigtran mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/sigtran