[AG-TECH] IGMPv3, Source Specific Multicast, fully supported in Windows XP

Fri Aug 16 13:06:14 CDT 2002

> So if you're auto-joining every source in a group, what benefit are you
> getting out of (S,G) versus (*,G)? 

Here are a few thoughts:

- From an infrastructure perspective, you are reducing complexity...
  no Rendezvous Point, and therefore no need to initially establish a
  shared multicast tree to share traffic before moving to a more
  efficient source-based tree.
- SSM for inter-domain multicast eliminates the problems with MSDP
  (see Bill's e-mail).
- Everyone auto-joining every source in a group may not be the best
  case to consider for SSM... missing the benefits of being able to 
  purposefully AVOID someone's source.

Jeff

Jay Beavers wrote:
> 
> So if you're auto-joining every source in a group, what benefit are you
> getting out of (S,G) versus (*,G)?  Are you assuming some UI to manually
> prune your (S,G) list?
> 
>  - jcb
> 
> -----Original Message-----
> From: Jeffrey Eschbach [mailto:eschbach at motorola.com]
> Sent: Friday, August 16, 2002 8:36 AM
> To: judson at mcs.anl.gov; ag-tech at accessgrid.org
> Subject: Re: [AG-TECH] IGMPv3, Source Specific Multicast, fully
> supported in Windows XP
> 
> > In the case of OM vic, this is already possible, fwiw.  The OpenMASH
> > folks have been exploring ssm for awhile.
> 
> Ya, but I think that Vic implementation only supports joining a single
> (S,G) pair.  Still need a way to specify a LIST of (S,G) pairs or even a
> list of sources for a single group.
> 
> > do they address the problem of distributing the sending sources
> > around? or is it at the level of giving vic a list of addresses.
> 
> You just give Vic the (S,G) pair.
> 
> My personal wishlist:  Before Vic distributes the sources around, I
> would love to see an implementation that allows dynamic addition/removal
> of (S,G) pairs.  Right now with (*,G) m-cast we are used to people
> joining/leaving a group at will without any trouble.  With SSM Vic and
> no dynamic add/remove, users would relaunch Vic every time they want to
> add a new (S,G) pair...  from an AG node perspective, imaging everyone
> in a venue having to relaunch Vic every time a new node joined.
> 
> Jeff
> 
> "Ivan R. Judson" wrote:
> >
> > In the case of OM vic, this is already possible, fwiw.  The OpenMASH
> > folks have been exploring ssm for awhile.
> >
> > --Ivan
> >
> > ..........
> > Ivan R. Judson .~. http://www.mcs.anl.gov/~judson
> > Futures Laboratory .~.  630 252 0920
> > Argonne National Laboratory .~. 630 252 6424 Fax
> >
> >
> > > -----Original Message-----
> > > From: owner-ag-tech at mcs.anl.gov
> > > [mailto:owner-ag-tech at mcs.anl.gov] On Behalf Of Tony Rimovsky
> > > Sent: Friday, August 16, 2002 8:24 AM
> > > To: Jay Beavers
> > > Cc: ag-tech at accessgrid.org
> > > Subject: Re: [AG-TECH] IGMPv3, Source Specific Multicast,
> > > fully supported in Windows XP
> > >
> > >
> > > In addition to the routing and forwarding issues that Bill
> > > mentions, there are application changes.  For example, vic
> > > and rat need to be able to say "i only want traffic from
> > > <list of sources> in <group> " or "i only want traffic from
> > > <list of source,group pairs>".
> > >
> > >
> > > On Fri, Aug 16, 2002 at 12:36:03AM -0500, Bill Nickless wrote:
> > > > At 10:01 PM 8/15/2002 -0700, Jay Beavers wrote:
> > > >
> > > > >Since I'm not a router bithead, can anyone give me a heads up if
> I
> > > > >should
> > > > >expect SSM to "just work" between Internet2 nodes?  Or
> > > might there be
> > > > >complications?
> > > >
> > > > Hi Jay,
> > > >
> > > > To answer your specific question: the Internet2 Abilene network is
> > > > ready
> > > > and capable of running SSM, and has been for well over a year.
> > > >
> > > > That being said, the problem with SSM deployment revolves
> > > around the
> > > > need
> > > > for network edge devices like switches and routers to
> > > support IGMPv3.  It's
> > > > difficult to find Ethernet switches that support IGMPv3,
> > > and they're not
> > > > the most economical choices.
> > > >
> > > > In a strange inversion of the normal situation, the backbone is
> > > > capable of
> > > > SSM service well before the local area networks of the
> > > connected sites.  If
> > > > you can find a site that supports SSM/IGMPv3 locally, it's
> > > a pretty good
> > > > bet that they can do SSM with most all other similar sites.
> > > >
> > > > ===
> > > > 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
> > > nickless at mcs.anl.gov
> > >
> 
> --
> ----------------------------------------------------------------
> Jeff Eschbach                              eschbach at motorola.com
> Networks and Infrastructure Research
> Senior Staff Engineer, Motorola Labs Internet2 Initiative
> http://internet2.motlabs.com              Fax:    (847) 576-3240
> Desk: (847) 538-5846                      Mobile: (847) 980-2240

-- 
----------------------------------------------------------------
Jeff Eschbach                              eschbach at motorola.com
Networks and Infrastructure Research
Senior Staff Engineer, Motorola Labs Internet2 Initiative
http://internet2.motlabs.com              Fax:    (847) 576-3240
Desk: (847) 538-5846                      Mobile: (847) 980-2240