Donnerstag, 17. November 2011

Performance ASM Redundancy Normal vs. Rendundancy External

Recently, I did do some performance testing and I want to share my results:

during no heavy workload periods
  • Swingbench Business transactions experience 56% more DB Time when switching to Normal Redundancy 
  • db file scattered read raised by 8% to 79%
  • log file parallel write raised by 10% to 49%
  • db file parallel write raised by 16% to 30% 
during heavy workload periods
  • Swingbench Business transactions experience 165% more DB Time when switching to Normal Redundancy
  • db file scattered read raised by 4% to 76%
  • log file parallel write shrinked by 5% to 49%
  • db file parallel write raised by 7% to 28%
The rule of thumb which we can derive:
  • If you have a system with heavy workload than you should not use Redundancy Normal.
Values for physical write bytes per transaction did not change.
This means DBWR statistics do not care about diskgroup redundancy.

Keine Kommentare:

Kommentar veröffentlichen