RE: Lossless fail-over

Jan,

O.K., this time I understood what you mean. A few points:
a) Relying on a single router for IP connectivity is really very very bad
network engineering practice for SIGTRAN deployments. Especially in the core
of IP network, I hardly can imagine this happening, i.e. that failure of a
single router causes two IP endpoints not to send packets to each other.
Please note that if there is an alternate path, SCTP will retransmit lost
packets.

b) If one uses a single router to connect to to the IP core (relying on a
single router for SIGTRAN purposes to connecto to the IP core is again a
very bad idea), failure of that router would isolate all ASPs in the same
subnet, e.g. if ASP2 and ASP3 in your example are in the same subnet, both
of them won't be able to send/receive IP packets to/from ASP1. If ASP2 and
ASP3 are in different subnets, they won't be able to synchronize among
themselves, because ASP2 can't talk to IP core.

We can analyze specific network configurations more and it probably is
possible to come up with some really interesting -and not realistic-
configurations where failure of a single router causes problems but I
believe the point is:
Failure of a single router shouldn't cause loss of IP connectivity.

     Thanks,
     Tolga

> -----Original Message-----
> From: Kolomaznik Jan [mailto:[email protected]]
> Sent: Monday, June 19, 2006 10:45 AM
> To: [email protected]; Tolga Asveren
> Cc: [email protected]
> Subject: RE: [Sigtran] Lossless fail-over
>
>
> My meaning is that classical SS7 "backup" solution is "equal" to the
> SCTP multihoming solution. Like in the SS7 network also in IP network it
> may happen that all signalling links may meet together on one STP (all
> mulhoming links on one IP router). Backup on COID level is another level
> of backup.
>
> Jan
>
> -----Original Message-----
> From: Brian F. G. Bidulock [mailto:[email protected]]
> Sent: Monday, June 19, 2006 4:19 PM
> To: Tolga Asveren
> Cc: [email protected]
> Subject: Re: [Sigtran] Lossless fail-over
>
> Tolga,
>
> Tolga Asveren wrote:
> (Mon, 19 Jun 2006 09:46:09)
> >
> > Loss of communication happens due to the crash of ASP1, because
> network
> > failures are handled by SCTP multihoming -as said this is the
> preassumption
> > which must hold with a very high probability. To discuss its validity
> is
> > another issue-.
>
> This could only be valid in a world without back-hoes.
>
> --brian
>
> --
> Brian F. G. Bidulock
> [email protected]
> http://www.openss7.org/
>
> _______________________________________________
> Sigtran mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/sigtran
>
> _______________________________________________
> Sigtran mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/sigtran