Networker and Exchange 2013 DAG - Performance issues leading to DAG failover

After 6 months of nearly perfect backups my Networker and Exchange 2013 environment have decided to cause me some headaches.

The environment:

Networker 8.1.1.2

NMM 3.0.1

3 Exchange 2013 CU5 MBX servers

6 DBs (all between 25-30GB), 3 active on each primary server and the last server is completely passive

I have my backup set to grab all of the DBs (3 active, 3 passive) off of one server.  This has been running great until a few days ago when the backup started causing databases to failover during the backup period, which caused the backup to fail in the end.  Checking the application logs I found all kinds of errors about ESE causing latency issues.  To fix that, I changed a few of the switches in my client from

NSR_ESE_UTIL_SEQUENTIAL=False

NSR_ESE_UTIL_THROTTLE=False

To

NSR_ESE_UTIL_SEQUENTIAL=True

NSR_ESE_UTIL_THROTTLE=True

NSR_ESE_THROTTLE_IOS=500

NSR_ESE_THROTTLE_DURATION=1000

Instantly corrected my ESE issues.  Now the backups are continuing to progress and get past the initial ESE check without failing any DBs over.  That being said, I am still seeing serious complaints from Exchange (and occasional fail overs) during the actual copy of the DBs and logs over to my Data Domain.  I tried setting the Client Parallelism to 4 to limit the number of save streams, but it continues to push 12 streams at a time, which I think may be part of the issue.  I want to limit this to 4 just to see if it will assist, otherwise I may have to change it to backing up the passive instances of each of the my DBs.

What am I missing in limiting the number of streams for a single client?  Limit it by volume?