DDBMA (Data Domain Boost for Microsoft Applications)

I recently found the existence of the DDBMA and was wondering the benefit of using it. We use DDBoost in the traditional way like this:

Have an MTREE with a CIFS share that is also replicated to an alternate remote DD unit. On SQL server we have a maintenance plan that dumps the information to a local drive on the server and it is then copied to the local DD CIFS unit which will then replicate eventually to the remote.  Having DDBoost on the units allows us to take advantage of faster performance to the DD's and for each backup the whole file is transmitted from the SQL server local drive to the local DD CIFS. It is my understanding that the nature of the DD's files are deduped in flight so while the whole file is being transmitted across the wire, the DD is smart enough to only really write the changes from the blocks.

How is this different when using DDBMA? My guess is that when using DDBMA the agent takes care of the comparison before sending the changed blocks to the local DD and thus saving the DD from doing the work. Is this really how it works?

Has anyone any examples of before and after using DDBMA as far as times and throughput that can be shared?

Also, if we have purchased DDBoost, do we also need to purchase DDBMA or that included in DDBoost and I just never knew about it?

Thanks

Paul