[AG-TECH] Multicast beacon questions

[Tony Rimovsky]

Hi, David.  I have recently taken responsibility for the development
of the multicast beacon in the NLANR/DAST group.

I understand your concern about beacon scalability.  We will be
examining mechanisms such as using RTP data in order to try to lower
the overhead of using the beacon in large communities.  

It is also significant to note that, for access grid Venues with large
numbers of senders such as the Lobby (224.2.177.155), you can see much
higher packet rates for the group.  There are 37 multicast sources in
the Lobby right now, and packet rates for the sources currently range
from 0 - 270 pps.  Eyeballing rates in that group, it looks like
active senders are averaging around 300 Kbit/s.  The 270 pps source is
generating ~2 Mbit/s.  You probably want to make sure your networking
folks know that the AG does generate a bit of traffic.  I think the
planning number is 1 Mbit/s per active site in a Venue.

By comparison, each beacon is sending ~5 Kbit/s into the beacon group.
The aggregate utilization of beacon traffic for the access grid beacon
isn't much more than the range of a single video source, even at
todays high level of usage w/in the access grid community.


Tony Rimovsky
National Center for Supercomputing Applications
Director, Networking Division
tony at ncsa.edu



On Wed, Oct 24, 2001 at 02:57:11PM -0400, David E. Bernholdt wrote:
> I think the first concern was security -- were we under some kind of
> attack.  After that, I think the question is whether that level of
> traffic is really necessary, since this is baseline (our AG node is
> not here yet) -- the beacon stuff is not a huge volume of data, but
> there are many packets, and that's taxing in its own way.
> 
> And if the answer is simply "yes, it _is_ necessary", then I think its
> okay. But it would be nice to be able to tell them that as the number
> of AGs builds, someone's working on a better way of monitoring things.
> --
> David E. Bernholdt                   |   Email: bernholdtde at ornl.gov 
> Oak Ridge National Laboratory        |   Phone: +1 (865) 574 3147    
> http://www.csm.ornl.gov/~bernhold/   |   Fax:   +1 (865) 574 0680