Behavior change between BE 2010 and 2010 R2/R3 unacceptable
Originally when we purchased BE 12.5 with the SSO option, resource volumes seen as local drives by the clustered servers were also seen as local drives by BE installed on those servers with the SSO license. So all that wasn needed was a backup job on that server to include all drives that the server owned. This behavior contin ued with BE 2010. THis was a GOOD thing. It meant that all HUGE volumes were backed up through the SAN fabric (fiber) and not across the production network. We liked that. Then with BE 2010 R2 all of a sudden, BE no longer saw san mounted volumes as local drives. We were told we needed to purchase the CASO option and the main media server would determine which of the clustered MMS servers owned the volume resources and delegate the jobs to the correct one (the one that owns the resource) which it did. Then we had a a problem and were told that BE 2010 R3 would solve all our problems with this particular issue. We upgraded and once again the behavior of BE changed. Now instead of sending the job to the MMS server that owns the resource, it will send it to which ever server it determines by some 'fuzzy math' is the least busy. Translation: it NEVER picks the right server to delegate the job to. I've tested this until my fingers cramped from all the keystrokes and mouse clcks. Not once out of 15 tests in a row did it pick the right server to send the job to. Since it is impractical to modify a job every time a resource moves from one server to another and it is also unresonable to expect us to accept that we can now only backup data at 1/4 the rate we used to after paying thousands of dollars for a solution - two solutions actually - that were to back it up at the higher rate.
It would be nice to have the logic written back into the algorythm that choses the MMS server to discriminate based on which MMS server ownes the resource instead of shich server it "thinks" is the least busy - which doesn't make ANY SENSE at all since none of them are used to backup other servers and consequently none of them are ever busy. The following diagrams illustrate how BE currently behaves and the way it used to behave - the way I'd like it to behave again. Thanks for yor consideration of my idea.