AWR Report issues

From Dirty Cache Wiki
Jump to navigation Jump to search

An overview of problems with manual data collection from database hosts which led me to write Dbcollect.



AWR can be produced in text or html format. There are versions for RAC and per-instance.


The current "locale" settings can have an effect on how timestamps are presented


Sometimes people produce AWR reports with very long intervals/time periods (one day or even much more). These are only barely useful. Like asking why there is a traffic jam on Mondays at 9AM where all data you have is the average amount of cars per hour for a full month.


"Our database size is 5 Terabyte". OK cool. Is this:

  • Size of tablespaces? Tables? Something else?
  • Size of ASM diskgroups or filesystems?
  • Allocated or used?
  • Including archivelogs, backups, exports, etc? Can we split by category?
  • Marketing or Engineering numbers?
  • Including snapshots Data Guard standby?
  • ASM / LVM mirrored copies?
  • Using compression? What ratio?
  • What is the retention of archivelogs, backups, etc?


Are more databases running on the same host? Are there other application processes?

System configuration

What CPU type and model is being used? Are we virtualized? What hypervisor? What OS level? Memory configuration? File systems? Attached disks?