Hello Norman,

1. to build a sync mirror with the server, then you can "set no failover" on primary" and "play" withe the backup, you could take IRIS down in any time on the backup, it will re-sync when its up again

2. O/S defragmentation - either by O/S tool, or to copy the DB to a new clean disk

3. Internal defragmentation - there are some options:
A) to copy all data with GBLOCKCOPY to a new DB, with 1 process, to keep data as sequential as possible (will take a lot of time)
B) to use the internal defragmentation tool (will need 2 x space on that DB) then compact globals

4. In general, its a good practice to split indices to a different DB (and use mapping) since the data ("D" globals) are usually not getting so fragmented over time. This will also help on future defragmentation, compact globals, DB "shrink" and other maintenance on less volume 

Hello Alex,

I would suggest using MONLBL with all measures, to better understand any "bottlenecks" in code, memory issues or global access. In parallel, to monitor the VM to see if we hit CPU, disk I/O or memory thresholds there.
it might be that there are some "internal" functions that can be optimized or written definitely to speed up code.
I recommend that you open a WRC, since they have good tools to analyze almost every aspect of IRIS

Thanks David,

The issue is that tasks are flag as "Reschedule task after system restart" = Yes, but IRIS was not restarted.
Only a mirror switch occurred. I;m not sure if a mirror switch does this re-calculation on times.