All things related to multicast

[Tony Rimovsky <>]

>From a beacon perspective, I'd be interested in what sort of time
intervals for testing people think would be appropriate.  

As part of the redesign we are looking at this and trying to strike a
balance between granularity and capacity concerns.

One option is the server being able to send commands to the beacons do
initiate a rate change for a period of time.  [Something similar to
this will be used to trigger group changes for beacon communities in
order to check for state building problems.]

Also, see <http://farmer.ncsa.uiuc.edu/~tony/beacon.html>.  This is
something I am playing with.  The interval for these plots is 120
seconds.  Is a view like this but with a much finer time granularity
valuable in the context of this discussion?


On Mon, Mar 04, 2002 at 01:57:01PM -0500, Marshall Eubanks wrote:

>     MTRG is hopelessly too coarse grained to detect most problems of 
> interest. We look at things from few second averages down to the audio
> or video frame level.
>