RE: multicast-transition (fwd)

[Hitoshi Asaeda <asaeda@yamato.ibm.com>]

Doron,

Thank you for reply.

> The protocol you are suggesting is a totally new multicast protocol. It is
> self-contained, i.e., does not need different join/leave protocol, and can
> automatically create tunnels. This protocol uses the "8-byte" notation,
> i.e., (S,G).

It may be seen it's a new. And your guess seems right.
But even you feel it's new, it'd be good for us if it'll be deployed,
wouldn't it?

> There was another protocol, with these specifications, called Simple
> Multicast. From various reasons it didn't pass the IETF. At this point, I

I'm not familier with the spec of Simple Multicast, but I know it.
If my idea is similar to Simple Multicast, that should be called
"Simple Multicast-SSM"? :)
Anyway, I'll check the detail spec.

> would say, there is no need for another multicast protocol. We have PIM-SSM
> / IGMPv3, which I think is ready, partially deployed, and will scale well to

partially deployed, yes. scale well is yes?
To scale well, MSDP was proposed, as you know, even MSDP itself may not 
be scalable.
But, we should not start PIM/MSDP's scalability issue here.

> the entire Internet. We should, however, fix some problems: tunneling,
> reverse-IGMP, and most important -- deploy it, and use it.

Totally agree.

> For the protocol itself, I think that it has a severe security bug: the
> forward tree setup. Other multicast protocols uses the (sub-optimal) reverse
> tree setup. With forward tree setup, there can be an attack from the
> multicast transmitter; it can send join packets to everyone in the Internet,
> and cause a flood. Any forward tree setup is subject to this attack. These

Yes, that's right. So, I wrote "there are some immature/unsolved
points, and one is security.".

> protocols will work only if you PAY per receiver; in this case, I am not
> sure that we need multicast at all.

I don't make an objection for your idea and my idea, mine is not
finished. But, I for one wish multicast is easiliy deployed step by
step.

Thank you.
--
Hitoshi Asaeda