Hi Sgt_why,
Usually when I see a lot of small images I will like them to be bundle in large Duplication jobs, this way you will experience less mounts and dismounts requests and NBRB will do better. Based on the parameters you have I will change it to:
MIN_GB_SIZE_PER_DUPLICATION_JOB = 200
MAX_GB_SIZE_PER_DUPLICATION_JOB = 800
MAX_MINUTES_TIL_FORCE_SMALL_DUPLICATION_JOB = 240
DUPLICATION_SESSION_INTERVAL_MINUTES = 60
This will allow EMM and NBSTSERV to better process and bundle those tiny images in to a big job, the trick here is if those images can be bundle because there are several rules behind the bundling like retention, destination, SLP policy and some other small factors, in my experience with backlogs troubleshooting is more an art than an exact science and you have to be very patient, do one change a day only and wait to see how your perfomance improves, then next day you will know excactly were you are, it will take a week or two to finaly get a perfect tuneup, I know is crazy but is what I have learn over all this years of dealing with the backlog beast.
If the changes I recommended dont work dont get disapointed, it will mean you have a lot of images with different retentions and destinations and you will need to tweak your SLP's config to send data to the same STU's or SUG's.
Hope this helps and let us know how it goes.
Best Regards.
This is are my personal opinions and of the company I work for.