Problem(Abstract)
A couple of simple settings that can be used by a ProtecTIER system administrator to help with excess fragmentation or pending deletes, on system without enough idle time.
Symptom
Fragmented space is increasing over time, even though your system is running at steady state. Or you have done a lot of special activities that resulted in more then usual deletes, and your system cant catch up on the extra fragmentation or deletes
Cause
The system isn't working in IDLE state enough, or the system isn't doing enough work, during busy cycles.
Environment
ONLY applicable to ProtecTIER versions greater the 2.5.x.x
Diagnosing the problem
It is important to have a feel for the trend, so that decisions can be made regarding how much change is needed. If the fragmentation is growing slightly, catching it early can sometimes be the best thing. However if its gotten far behind over a long period of time, it may be better to help it catch up, then adjust it again to prevent it from falling behind again.
Resolving the problem
This technote talks about 2 approaches to modifying MRC. There are other approaches that can be used, and they can all be combined. But for simplicity, if we are able to solve the problem with these 2 approaches, in the long run, the system will be easier to administer.
Approach 1:
View the Statistics in the ProtecTIER manager and see if there is a pattern to the workload (over the entire cycle, that will allow you to slightly increase the "IDLE THRESHOLD" and have your system working on the Fragmentation and Deletes enough to keep up.
1.) Check what your Idlethreshold currently is :
# /opt/dtc/app/utils/vtlutil -rpc GetIdlenessThreshold
2.) Looking at your statistics in the Manager Screens, if you have some pattern like this:
and your current idlethreshold is 25 - you would adjust your idleness threshold to 45 or 50, to allow it to work in the lower vally where the system is running less then 50 MBs second.
3.) Adjust your idlnessthreshold:
#/opt/dtc/app/utils/vtlutil -rpc ChangeIdlenessThreshold -int 45
4.) Monitor your system to make sure performance is still ok and that you are seeing your fragmentation decline
Approach 2:
View the statistics screen in the ProtecTIER manager and see if there is a pattern to the workload that shows your system will be in the BUSY state all day long, or for the majority of the day. Then verify if there is extra space between the Maximum planned throughput and the normal activity. If so, then the disk utilizations can be adjusted slightly, to allow the system to do more work even when it is BUSY.
1.) Check what your disk utilizations currently are:
#/opt/dtc/app/utils/vtlutil -rpc GetMaintenanceDiskUtilizations
2.) Increase the 3 BUSY values slightly.
#/opt/dtc/app/utils/vtlutil -rpc ChangeMaintenanceDiskUtilizations -int 230 -int 20 -int 28 -int 37
Note, there usually is no need to change the IDLE from 230 at this stage. But if the three values were 15, 25,35 before (the default), then a slight adjustment to 20, 28, 37 may be all you need to allow it to work just a little harder. You dont want to adjust these too high, as they will impact your performance for your backup/restore and replication workload.
Conclusion
Be sure to monitor the affects of your changes, and save your original values so that you can easily set them back. These values can be changed without any disruption, but it will take at time to notice the affects of these changes.
If you are unable to find a suitable change that helps, please call IBM SW support for additional recommendations.
