Re: Questions on session level multicast simulations

[Polly Huang <huang@catarina.usc.edu>]

On Tue, 7 Jul 1998, Shuqian Yan wrote:

> Hi,
> 
> Maybe Polly Huang from USC/ISI can help answer my questions on the session level 
> simulations. 
> 
> (1) From the sample tcl script on session level simulations, there is no 
> specific multicast protocol indicated in the script. How come?


1. We want to leave out as much detail as possible (ie, almost no routing 
mechanism at all)
2. We want to have the most simple form of packet 
distribution (ie, one-to-one connection is a special case of one-to-many)
3. To evaluate of session level simulations, we consider the worst case, 
dense mode multicast (source specific trees)

Keep in mind that one session is not equal to one multicast group.
Each session is formed by one root and one or many receivers.
If you'd like to simulate shared tree or multiple source groups using the 
session level simulator, you could create sessions from the
sources of the shared tree to the rendezvous points, and then create 
sessions from the rendezvous points to the receivers.  

However, at this point, no simple API is available for multiple 
multicast sources or shared tree multicast simulations. Sorry.

> (2) For session level simulation, is there any limit on the number of sources 
> per session? 

As mentioned above, 1 'session source' per session. However, again as 
mentioned above, multiple 'multicast sources' can be simulated by having 
multple sessions.

Hope this answers your questions.
-Polly

> 
> Tks.
> 
> shuqian
> 
>