All things related to multicast

[Bill Nickless <>]

At 03:09 PM 6/12/2001 -0700, Toerless Eckert wrote:
> What's the difference between (*,*,*) switching as you call
> it and effectively shared media ? If you only have routers, and that's
> what we're talking about in a MIX< it is effectively a shared media. You
> won't be able to squeeze more bits out of it except for a little bit
> better buffering and avoidance of collision detection.

"effectively shared media" is what you get when you naively build a layer 
2  exchange point out of an Ethernet switch and people start using it to 
exchange multicast traffic.

(*,*,*) switching is what you get when you turn on IGMP snooping on that 
same exchange point because IGMP snooping detects only the 
multicast-capable routers.  (Alternatively, you use a separate VLAN for 
multicast.)  Now the unicast-only participants don't get the multicast traffic.

> I think that what you call (S,G,RPF) switching is not for what
> you say: elimination of duplication (that already should work in the
> other solutions), but instead it is about the ability of each downstream
> peer to individually select it's upstream peer for a certain traffic
> source. This of course doesn't apply to the last case you cite, the
> L3-MIX. In addition, it is i think orthogonal to the mechanisms described
> above, eg: in the solution i proposed, you could use RGMP or PIM
> Snooping in each VLAN too.

I don't think the two reasons (dup elimination, upstream (s,g) choice) are 
orthogonal.  You get the duplicates WHEN some participant chooses the 
"wrong" peer to send the PIM-SM join--that is, they pick a different RPF 
for an (S,G).  You want to support the ability of participants to choose 
their RPF neighbor for reasons that include the avoidance of duplicates.

Thinking about it more carefully, your solution isn't quite fully (S,G,RPF) 
even with RGMP and current PIM-SM snooping.  Consider two sources S1 and 
S2, and three participants A, B, and C.  A might want to get (S1,G) from B, 
but (S2,G) from C.  The exchange point would be forwarding (*,G) from both 
B and C, which means A could still get duplicates--(S2,G) from B or (S1,G) 
from A.

> Finally, if you want to make a taxonomy, then it might be interesting
> (although most likely not of much practical use), that the most
> efficient and flexible method to set up a MIX is to use RFC2337, eg:
> mapping of PIM to ATM Multipoint signalling. You use an ATM-switch
> with ATM-Multipoint signalling and replication and each peer who
> sources a (*,G) or (S,G) establishes an ATM point-to-multipoint SVC
> to those peers on the mix interested in that stream from exactly this
> upstream neighbor. So you have the full multicast efficiency and 
> flexibility: Each upstream peer only needs to send out data once, and each 
> downstream peer only receives traffic from that upstream neighbor that it 
> selected by it's local routing decisions. What i proposed with ethernet 
> and VLANs achieves the same advantages but is of course more ugly to set up
> - that's the price to pay for wanting to use inexpensive ethernet ;-))

Very good point.  I forgot about that one!  Just for my curiosity's sake, 
do you know who has that implemented, and if anyone is using it?

===
Bill Nickless    http://www.mcs.anl.gov/people/nickless      +1 630 252 7390
PGP:0E 0F 16 80 C5 B1 69 52 E1 44 1A A5 0E 1B 74 F7