North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

sparse vs dense and multicast

  • From: smd
  • Date: Tue Jun 15 19:56:38 1999

[email protected] wrote:

| and are you suggesting that dense-mode (blindly distributing), to coin the 
| term, is more efficient than say, sparse-mode (multicast?) .. not even in your 
You are misusing two reasonably well-known terms used in the IP multicast
world today.  You could easily argue that the PIM use of "sparse" and "dense"
is gross and in some way wrong, however they are understood to be very
different from dense = "blindly distributing" and "sparse" = 
"undefined but maybe multicast".

In particular, in PIM [DEFJLW 94] a sparse group is one in which only
a few widely-scattered receivers have joined a given group, while a
dense group is one in which most multicast routers are likely to have
at least one receiver joined to the group.

The sparseness vs denseness of a group helps one decide if
flood-and-prune is efficient or whether it is more efficient 
for an explicit join towards the root of the [local-]spanning 
tree to succeed before traffic is sent.

Doing the latter on a dense group obviously, can result in a join
implosion towards the [local] root of the spanning tree, while
doing the former on a sparse group results in the arrival of lots
of unnecessary traffic prior to the issuance of [lots of] prune

"dense-mode" and "sparse-mode" are PIM terms: with PIM any group 
can be declared sparse or dense.

The use of "ip pim sparse-mode" etc. as an interface command
in IOS is often confusing.  Making the terms worse by 
further overloading them is unhelpful.

| -danny (who won't waste any additional time in this thread)

That's probably wise.