2013-01-31 04:35 PM
I want to monitor my NetApp 6210 and 3170 SAN storage. I am pretty much done with everything. I am using open source third party NMS and doing SNMP based monitoring of my storage infrastructure. i am struck at below points:
1. Number of IOPs per volume.
2. Aggregate number of IOPs on all volumes.
3. Number of IOPs on each physical disk.
Prompt response in this regard will be highly appreciated.
2013-02-01 07:45 AM
You can use NetApp "stats" command which is available in the Data ONTAP on any filer.
You can find out the available measureable objects on the filer by entering:
toaster>stats list objects
As you are initerested in 'lun', 'volume' & 'aggr' related stats, you can see them listed among objects listed by the previous command.
To see 'counters' available for the objects listed in step1:
toaster>stats list counters volume
Counters for object name: volume
toaster>stats list counters lun
Counters for object name: lun
toaster> stats list counters aggregate
Counters for object name: aggregate
You can go one step further and determine what instances are available for the object; say ' volume'. The instances are basically the volumes available on that filer.
toaster>stats list instances volume
Instances for object name: volume
Now, Start system statistics gathering in the background using '-I' as identifier. For example: To observe stats for volume 'vol_iscsi_win_test':
toaster> stats start -I volstats volume:vol_iscsi_win_test
Note:volstats is just the name I have given, you can give any meaningful name as you like.
To see the results while I/O is in progress:
toaster> stats show -I volstats
Note:I am running these commands on my simulator, so dont read too much into the vaules here.
To stop the background stats and print the result on the console, enter:
toaster>stats stop -I volstats
Similarly for LUNS:
toaster> stats list instances lun
Instances for object name: lun
toaster>stats start -I lun_stats lun:/vol/vol_iscsi_win_test/iometer-BWKgW]BaIqNQ
toaster>stats show -I lun_stats
toaster>stats stop -I lun_stats
There is alreay a wonerful blog post on this matter:
Performance "stats" without PerfStat or Ops Mgr:
2013-02-01 09:15 AM
Thanks for your response.
What i want to do is. i want to have IOPs for each volume as separate and aggregate by using a third party NMS (Zabbix) via SNMP. I need to know the OID's and MIBs which i can download and can do the needful.
2013-02-01 11:43 AM
IOPS, and latency, etc are not available via snmp, only via the Ontap API.
Roll your own, or go with something that has this built in and automated. (e.g. http://www.logicmonitor.com/monitoring/storage/net
2013-02-04 08:45 AM
I can have IOPs per protocol level i.e. Number of IOPs on NFS, iSCSI and FC but i wanna have IOPs per volume level. I think if i can have at protocol level then there must be some way out for volume level as well.
2013-02-05 09:30 AM
You would think. But unfortunately it's not true. :-)
Having been through the NetApp MIBs comprehensivley several times, I can state categorically there is no way to get per volume IOps or latency.
Stuff like this:
can only be gotten via the API.
(The above is from logicmonitor).