Has anybody looked into the accuracy of tape drive throughput reports?
Report Templates > Disk & Tape Device Activity Reports > Tape Drive Throughput
A long running backup ran for over nine hours had six streams running at ~18 MB/s each into the same drive in parallel, total throughput is about 108MB/s. In the tape drive throughput that tape drive never exceeded about 5MB/s (actual 40716Kb/s) when grouped by "hour of the day", plus I would expect the report to show the tape drive and media server combination reporting throughput over an extended period. The period indicated in OpsCenter is substantially shorted than what NetBackup reports in the activity monitor. Usually were there is one obvious error in reporting there probably others. All the throughput rates are in Kb/s even if they were in KB/s that would not be close to reality.
According to the manual they are calculated on "Advanced averaging logic is used to determine the throughput values for each cell in the report. The averaging is done at an image fragment level ensuring granularity and precision in calculations." - marketing ?
Just a note on the tape drive throughput report - in the event you run it for last 24 hours it will report until "1 AM last night" - this because the data collected is inserted into some temp tables and then at 1 AM every night there is a Sybase database event which copies this collected data to "permanent" tables (well, not permamanent but rather data will stay according to purge settings). And then the report will fetch data from these tables.
Marketing. Never been convinced by Opscentre and throughput , never mind drive hotspot/occupancy. I wrote my own occupancy code; some robotics management software eg slconsole now provide these data (occupancy) which do a much better job of it.