Re: [pim] Re: BSM: is it bidir or ASM

[Stig Venaas <stig.venaas at uninett.no>]

Nidhi Bhaskar wrote:
> > > I'll discuss with other authors and get back to you. The options seem
> > > to be to either explicitly make a decision to include only one range
> > > at the BSR or see if we can find a backwards compatible way to include
> > > both...
> >
> > Great, one other point for you to consider is how things can get messed
> > up on the BSR itself.
> >
> > The BSR is responsible for tracking all the RP-candidates and so it would
> > know that one RP was bidir while the other was ASM and so it may refer
> > to its RP-cache and find a different RP than the BSR clients would.
> > (This might be an implementation question, but I'm pretty sure that
> > most implementations will have this problem)
>
>
> Yes, I think this is an implementation issue.
>
> The spec already states that the BSM may contain only a subset of the
> mapping cache available at the BSR. We could add in some text to
> indicate that when this is the case the BSR should only consider this
> subset to determine the groups mode/RP. But it seems obvious...

The BSR is generally free to choose from the different C-RP-advs what 
group ranges and RPs to use for the BSMs. The BSR may then also sort out 
possible conflicts regarding bidir/sparse. Whether the BSR prefer bidir 
or not don't necessarily need to be standardised. It could even be 
configurable. OTOH if the BSR just announced the ranges and left to the 
routers to make a decision, then all the routers would have to always 
make the same decision. So my suggestion is to just add a sentence or 
two, making it clear that the BSR is free to form the RP-set, and that 
it should resolve such conflicts.

Stig

> Nidhi.
>
> _______________________________________________
> pim mailing list
> pim at ietf.org
> https://www1.ietf.org/mailman/listinfo/pim


_______________________________________________
pim mailing list
pim at ietf.org
https://www1.ietf.org/mailman/listinfo/pim