Hello,
I’m running a stress test on a RHEL 2.6.32 (256 GB ram / RAID 0 SSD ~300GB / Xeon E5-2670 2.60GHz 2 CPU 8 Cores Hyper-Threading)
ASE 16.0 GA PL05 configured with 16K page size using a default pool of 16 thread is configured for 36 GB of memory is using Huge pages.
All data can fit in the caches.
The load injector send ~40 parallel requests to an Open Server which then make ~20 parallel requests to the ASE.
The result on the sp_sysmon raise at least these 2 questions for which I need your help:
- What can be done to reduce Minor Page Faults at OS as for me it’s pretty high considering the HW (~600pf/s) and we already use huge pages?
- Do I need to bother to have a Task Context Switches Due To: Network Packet Received ~ 80 % ? Is there a miss network configuration as Tuning Recommendations suggest ?
Is NetController Activity / Polls Returning Events ratio of ~20% 'normal' ? I’ve tried to add more NetController without success.
Any other recommendations ?
Thanks in advance
F++
PS: I’ve attached a sp_sysmon of 1 min taken during one of the stress tests.
