1 media server per every 100 clients is enough and for clients with more then 1TB you can configured them has media servers and assign them a couple of drives, can be dedicated drives or shared, with 48 drives on your library the question here is how much storage are you backing up? and what kind of drives do u have LTO2, LTO3?
Im sure you are having a bunch of 196's, your first thing to do can be, review your policies and check if you have clients around multiple policies so you can prevent repeated runs, then disable any test or dev policy you have, only leave the production ones alive, then remove every Cross Mounted Points that you have configured under any policy and ensure that your are backing up those Mounted File Systems directly from the device where they are stored.
Check the total memory you have on your media servers and based on the number of drives on the STU, multiplexing and multiple streams that run under this media, you can determine if you are overloading the media server or if you can add more drives to the STU
Then go to the activity monitor and sort the data by total storage backed up and see which clients are the biggest ones and how long they are taking, if you u are using only one stream to back them up, you will need to reconfigure your policy to be able to use multiple streams and multiplexing (4 max), so you can run multiple jobs for the same client at the same time, reducing the backup time.
Check how your buffers are configured under your actual media servers and reconfigure them to be:
NUMBER_DATA_BUFFERS = 128
SIZE_DATA_BUFFERS = 256MB
NET_DATA_BUFFERS = 256MB
and configure the NET_DATA_BUFFER to 256MB on every single client you have.
Check for a couple of days how this works and let us know if you still have issues.
Hope this helps.
regards
