For folks who have been trying to do multicast deployment, the limitations of the MBONE have been excentuated.
The DVMRP part of the MBONE is a source of a lot of problems.
Some folks believe that individual backbones will have their own services that may not interoperate. This runs contary to the central spirit of the Internet.
There is no standard Multicast IGP.
There is a specification for extending BGP to carry protocol information. This new protocol is called BGP+. Using BGP+ for multicast is now being called MBGP.
BGMP is another potential solution, but it is still early in its development cycle.
DVMRP does not have the concept of peering. All connections are transit connections.
Some metric redistributions between PIM and DVMRP may make the metrics useless.
Loop can easily develop if a site is connected to two different parts of the MBONE.
See a web page for some of the protocol interactions.
A new architecture:
There are two testing beds for doing multicasting peering at Sprint and PAIX. They are not running yet,but real be happening soon.
Run sparse mode backbones and dense mode interconnnects.
There needs to be a concept of a border router to permit the sparse mode RP to exist correctly.
To fix this, the testing beds will work to resolve this.