RE: A better name than ISM

["Michael Luby" <luby@digitalfountain.com>]

Well, I view SSM as somewhere in between IP unicast and the original IP
multicast model, and in fact I think of SSM as a natural interpolation
between the two that has a lot of advantages over IP unicast and the
original IP multicast model, as well as some of its own natural limitations.
Thus, I'm not understanding why it is a very big departure from the entire
IP world.

It seems to me that unicast has the most explicit association:
sender is explicitly named and not changable.
receiver is explicitly named and not changable.

SSM is next:
sender is explicitly named and not changable.
receiver set can vary dynamically.

Then, the original multicast model:
sender set can vary dynamically.
receiver set can vary dynamically.

And, yes, it is ok to use the word fashion instead of compelling, if fashion
is defined to mean what people are *currently* putting a concerted effort
into to deploy commercially and create a revenue stream.  And you are right,
it may be fashionable in a couple of years to want to deploy commercially
applications that are more suitable to the original multicast model.  Let's
see.

I'm sure we can continue this conversation at a beer BOF ... (sunday
evening, monday evening?)

Mike

-----Original Message-----
From: Jon Crowcroft [mailto:J.Crowcroft@cs.ucl.ac.uk]
Sent: Sunday, November 26, 2000 7:21 AM
To: Mark Handley
Cc: ssm-interest@external.cisco.com; J.Crowcroft@cs.ucl.ac.uk
Subject: Re: A better name than ISM



right - ssm is not multicast. its really a very big departure from
the entire IP world -

the difference between unicast and deering's multicast at the
fundamental service level was very thin - anyone can send to either,
and anyone can receive what is sent to them,
or choose not -the rest was address allocation and routing

the channel model in SSM explicitly creates an association from
address allocation, source, and receiver set that is much less
flexible (although as everyone keeps saying, appears to cover the
majroity of the _current_ "compelling", or as i prefer, fashionable
applications:-)

i'd prefer we change the name of SSM back to channel

In message <26247.975114786@hazard.aciri.org>, Mark Handley typed:

 >>
 >>>At 12:32 PM 11/24/00, Hugh Holbrook wrote:
 >>>>I agree.  I think we should raise this issue at maddogs.  It feels
 >>>>kind of like a nit, but I think naming is really important -- bad
 >>>>names can cause SO much confusion...
 >>>
 >>>OK, there seems to be consensus that "Internet Standard Multicast"/ISM
is
 >>>not the best name/acronym for the current multicast model.  As for an
 >>>alternative - IMHO, either "Any-Source Multicast"/ASM or
 >>>"Source-Independent Multicast"/SIM would be better.  Personally, I
prefer
 >>>"SIM" for (as Mike noted) its symmetry with SSM (given that SSM stands
for
 >>>"Source-Specific Multicast", *not* "Single-Source Multicast".)
 >>
 >>SIM is a rather overloaded abbreviation, and ASM implies the receivers
 >>have no control over which sources they listen to, which is no longer
 >>true with IGMPv3.
 >>
 >>The key difference is that in SSM receivers join a source/channel,
 >>whereas in ISM they join a group (or group of sources).  The ideal
 >>name would probably be Group-Specific Multicast, but that acronym is
 >>taken.  Source-Group Multicast would work too, but (informally at
 >>least) that acronym is also taken.  How about Group-Addressed
 >>Multicast (GAM), Internet Group Multicast (IGM) or just Group
 >>Multicast (GM)?
 >>
 >>Cheers,
 >>	Mark
 >>
 >>

 cheers

   jon