Flowchart of performance data analysis
NetBackup performance tuning can be grouped into three major categories: H/W resource availability, Kernel/OS tuning, and application, i.e., NetBackup tunings. RCA and tuning should be conducted in that order.
Jobs consume hardware (H/W) resources, in particular, the four major H/W resources: CPU, memory, I/O and network. As NetBackup servers age and the workload changes over time, hardware resource usage patterns will change also. Therefore, regularly monitoring/checking for the H/W resource usage pattern is important and should be the first step of any performance issue troubleshooting. This is because if the server is bottlenecked on any H/W resource, attempts to tune NetBackup to make it run faster will be futile, since the bottleneck will throttle the workload and prevent any further performance improvement. Therefore, the first step of RCA of a performance issue should always be checking for the H/W resource availability.
Boxes 1, 2, 3 and 4 in the above flowchart summarize the steps to detect and remove any H/W resource bottleneck. Some H/W resource shortages can be alleviated by kernel tunings. For example, on the Linux platform, frequent heavy swapping in the hundreds and thousands of KB per second is an indication of a memory bottleneck, thus, lowering the kernel parameter, swappiness, can reduce swapping without adding additional memory.
After all the H/W or kernel resource bottlenecks are eliminated, and the performance problem still exists, then most likely the root cause is in NetBackup. RCA and tuning of NetBackup should be the next step. Boxes 5, 6, 7 and 8 are the steps for finding the root cause of NetBackup bottlenecks.