System object

[pure]

Hi,

I am trying to understand what exactly is a system object from AutoSupport Point of view. 

Does a 'system' mean [controller + RAM + CACHE(FP, FC) + DISKS] or [Controller + DISKS] or something else ? 

When you see IOPS - [nfs_ops, Cifs_ops ..] or [read_ops, write_ops, other_ops]  does it include ops being served from cache(if exists). 

I tried to match the ssd_data_read /write from system object with aggregate , but no match .

What exactly is system object ? what exact entities in storage solution is it trying to give you data for ?

is system object aggregated over all the volumes ?

I have read many articles, but I am kind of stuck on the fundamental thing.

Any pointer will be helpful. 

Thanks! 

[SpindleNinja]

You can view performance data for specific objects in your cluster. Objects are comprised of instances and counters. Counters provide data about the instances of an object.

An object is any of the following:

• Physical entities such as disks, processors, and ports
• Logical entities such as LUNs, volumes, and workloads
• Protocols such as CIFS, NFS, iSCSI, and FC

https://library.netapp.com/ecmdocs/ECMP1196798/html/GUID-5F8A4353-9A61-4A01-9D7C-7EBB1E0301E3.html

are you looking at RAW autosupport dumps? 

[pure]

Yes. And i did a bit of digging . 

My root problem in understanding is - will sum of ops on all volumes derived from volume object  be equal to ops  for that node from system object ?

[SpindleNinja]

not really.  There are meta data / wafl processes that will use IO too.   

[paul_stejskal]

Correct. It may not necessarily add up, as there are quite a few background operations which are not tracked. This KB explains why aggregate and volume won't add up in AIQUM, but the concept is similar: https://kb.netapp.com/app/answers/answer_view/a_id/1001962/loc/en_US#__highlight