[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Sparse Mode Multicast support (fwd)
some delivery problems, so forwarded again.
shuqian
---------- Forwarded message ----------
Date: Wed, 18 Nov 1998 16:33:15 -0500 (EST)
From: Shuqian Yan <[email protected]>
To: Deborah Estrin <[email protected]>
Cc: [email protected], [email protected]
Subject: Re: Sparse Mode Multicast support
Hello all,
Based on my simulation needs, I suggest the new sparse mode multicast module
should support the following basic features,
(1) Rooted shared tree at RP with explicit joins from individual receiver.
(2) able to switch to source based shortest path tree for a (S,G) pair if
the source data rate is high enough to trigger the switching from
shared-tree to source based tree.
(3) There should be explicit join and register messages simulated
(when one is concerned with overhead cost).
(4) To support bi-directional forwarding shared-tree and uni-directional
forwarding for source-based tree. Has this feature been included in the
new module? As my knowledge goes, the current multicast forwarding scheme
in NS2 does not support bi-direction forwarding. To completely change the
current forwarding model in NS2 is quite a tough task for me at the
moment.
thank you!
shuqian yan
On Mon, 16 Nov 1998, Deborah Estrin
wrote:
> We are in the process of putting in place some cleaner, more modular support
> for sparse mode multicast. We hope to have some of it merged into the
> current ns branch by next week sometime.
>
> Our thought was that most people using hte old PIM code would b well served
> by this new code, perhaps with some additional enhancements by us as we hear
> from the users.
>
> If you are a currrent PIM user we would like to hear which of the following
> features you need:
> * explicitly joined source trees
> * shared tree used for data only
> * shared tree used for rendezvous
> * etc.
>
> There are other issues to discuss related to which protocol details are
> actually simulated and we want to hear form users to prioritize our
> development.
>
>
>