Re: can ssm and non-ssm colide ? (re-sent)

[Marshall Eubanks <tme@21rst-century.com>]

Dany Shterman wrote:
> 
>     Hi.
> I understand that a router can have 2 types of multicast forwarding rules:
> 
> 1. (S,D) : i.e do multicast to a packet that arrives with IP destination
> "d" and IP source "s" iff S=="s" and D=="d". 2. (*,D) : i.e do multicast to
> a packet that arrives with IP destination   "d" iff D=="d".
> 
> Questions:
> a. is it true ?
> b. is it possible that a router will have 2 contradicting forwarding   rules
> : (S,D) and (*,D) that direct traffic to 2 different ports,   for the same
> IP destination address D ?
> c. If the answer to the previous question is true, then to what port
> should the router forward ?
> 
> ThankX.
> Dan
> 
> _____________________________________________________________________________________

Sorry - this was sent too soon - here is the final

Dear Dan;

What exactly do you mean ? Is D the multicast class D address ? This
is generally called "G", for the group. It is NOT a destination.

Remember, SSM and ISM address ranges are supposed to be disjoint - 
(S,G) is allowed in a SSM range, but (*,G) is supposed to be
filtered out. So this cannot occur in a proper SSM implementation. If  you
tried to do ISM in the SSM address space, it wouldn't work.

In PIM-SM ISM, 
a (*,G) JOIN and an (S,G) join certainly could have different directions -
generally, they would. It also is always possible to have the same
data arriving at a router from two directions - this is also supposed
to be detected and pruned.(SSM could have this too.)

In general, these situations are supposed to be detected and pruned
off as soon as possible.