Skip to Content.
Sympa Menu

wg-multicast - RE: IPv6 Multicast

Subject: All things related to multicast

List archive

RE: IPv6 Multicast


Chronological Thread 
  • From: "Michael Luby" <>
  • To: "Marshall Eubanks" <>, "Alan Crosswell" <>, "Kevin C. Almeroth" <>, <>
  • Cc: "Michael Luby" <>
  • Subject: RE: IPv6 Multicast
  • Date: Wed, 7 Aug 2002 09:38:31 -0700
  • Importance: Normal

I'm always very confused by these discussions, probably because of my bias
in terms of viewing multicast as just an underlying network efficiency
enhancement instead of some kind of service that an end user should ever be
aware of. However, given my bias, I am really perplexed why it is
considered the case that:

(1) for SSM, the perception is that a miracle has to occur to discover the
source address, but no similar miracle has to occur to discover the group
address (which also has to occur for ASM). There are millions of ways to
propagate the group address for ASM, or equivalently a (source, group
address) for SSM, including building apps like SAP, SDR, etc. if you like
that model, or webpages, etc.
(2) for many to many apps, people seem to think that ASM is ok. I can see
that SSM is more of a challenge in many people's minds for many-to-many
apps, but even there I can see some pretty easy ways at the app level to
support a many to many app using SSM that is at least as robust and
efficient as using the existing underlying ASM implementations.
(3) webpages and other standard mechanisms viewed as terrible ways to
disseminate info about content being delivered using multicast. These
mechanisms work quite well for unicast, and there are some extremely popular
websites out there that get tens or hundreds of millions of hits a day, so
why is this bad for multicast? Again, this may my bias that I consider
multicast an underlying network technology that should never even be exposed
to the end user. If you accept this bias, then there is fundamentally
nothing different between what the presentation to the end user would be for
multicast or unicast, and all the apps that need to be written on top of
unicast for reporting also would need to be written for multicast to support
reporting, ditto authentication, access, advertising the content, etc. So,
if there seems to be a need to generate new ways of advertising info to end
users for multicast, don't we need exactly these same new ways for unicast
(again, this reasoning is only valid if you buy my bias that multicast and
unicast should look the same to an app user, and multicast only provides
underlying network efficiencies over unicast).

Mike L

-----Original Message-----
From:

[mailto:]On
Behalf Of Marshall Eubanks
Sent: Wednesday, August 07, 2002 8:21 AM
To: Alan Crosswell; Kevin C. Almeroth
Cc:

Subject: Re: IPv6 Multicast


On Wed, 7 Aug 2002 10:09:54 EDT
Alan Crosswell
<>
wrote:
> Can someone explain to me how one uses SSM to do something intrinsically
> easy for ASM: discover the existence of unknown sources?
> /a
>

The general answer to this
is that you will discover SSM sources from a web site.

The discovery of unknown web sites is left as an exercise to the reader :)

Only people who

a.) do not care about the size of their audience or

b.) have very deep pockets

would be happy with this reasoning.

It seems clear to me that SSM will have SAP/SDR like guides, and I wonder
that
these have not been created yet.

Marshall





Archive powered by MHonArc 2.6.16.

Top of Page