Host os : 1+
4 weeks ago
According to this KB all these 'Domains' has CPU Concurrency of 1+ CPUs:
Host os : 1+
exempt : 1+
SM_Exempt : 1+ [No SnapMirror configured yet]
WAFL_XCleaner : 1+
WAFL_EX : 1+
ssan_exempt : 1+ [No FCP/iSCSI processing, Pure NFS environment]
XO_Ex : 1+
raid_exempt : 1+
storage : 1+
nwk_exempt: 1+ [IP & NFS Processing]
Given a situation, where all the above domains except (SM_Exempt & ssan_exempt) are running parallely and are able to process information using single CPU. So, total '7' CPU are being used at a given time.
For a NetApp FILER such as AFF-A300: 16 Cores [16-7=9]
My question : If need be, will 'nwk_exempt' domain be able to maximum it's processing using all the available CPU cores ? Or there is a limit to, it can grab ?, given they are indeed available.
@Alex Dawson : Do you mind attending to this query, you have been very helpful in the past on this subject.
Solved! See The Solution
3 REPLIES 3
3 weeks ago
This veers too far into our trade secrets to discuss publicly.
Please log a support ticket outlining your performance concern, provide me the ticket number, and I will ensure it gets dealt with appropriately.
Just wanted to add a bit of information here:
CSMP domain scheduling logic and domain grabbing is unique to ONTAP ,obviously that's the proprietary info. But, the following command is quite useful in understanding 'domain' busy/processor info:
set diag [ Under diag : more options show up for processor related counter]
statistics start -object processor -counter domain_busy -sample-id [1..100] example : 2
statistics stop -sample-id 2
statistics show -object processor -counter domain_busy