Re: Number of links towards a DPC (ITU)

"David Laight" <[email protected]> Tue, 18 Jun 2013 09:57:30 +0100
Newsgroups gmane.ietf.sigtran
Message-ID <[email protected]>
> Yeah, I got it wrong.
> My idea was to use the SLS but still generate enough SLS for the linkset locally.
> But this would cause the problem of mis-sequencing when links go down.
> But that seems to be the only other alternative of using the entire 32 links.

For locally generated traffic the local ISUP/SCCP could generate
more than 4 bits of sls (it probably generates 8 bits for ANSI
and mtp3 masks them for ITU) so MTP3 could loadshare over all
32 links - but your mtp3 is unlikely to behave that way.

If the traffic is TCAP/SCCP and 'route on GT' and you have
2 (or more) 'next hop' destination SCCP systems you can get
the TCAP traffic to use all 32 links.

Similarly 'class 0' SCCP traffic (order not guaranteed)
probably has an SLS assigned sequentially by SCCP.
The local MTP3 could safely reassign the SLS so that it
could use all 32 (or more) links in a combined linkset.

We have some other configurations that can loadshare over
32 links.

	David