2014-06-11 12:30 AM - last edited on 2014-12-02 12:52 PM by alissa
I have cluster mode controllers monitored with OCUM 6.2 (for clustered mode controllers) and OCPM 1.0. I want to pull out performance graphs like you could do with "NetApp Management Console".
How do I do it?
Thanks in advance.
2014-06-11 01:54 AM
If i am correct, You are asking for determining performance issues on clustered dataontap using OPM
In OPM1.0 we can view performance for individual volumes
On the below link check for section --> Determining whether a workload has a performance issue
We can check
- slow response time for a workload as well.
- Identifying victim and bully workloads
- we can determine QOS limit for high performing workload
2014-06-11 05:57 PM
Thanks for the information Srini. But I am after more detailed graphs. Something that you pull out of DFM 5.1 NetApp management console.
Graphs showing the "Per protocol IOPS", "Latency per protocol","CPU utilization per Node" etc.. I hope this give you an idea about what I am after.
2014-06-12 07:20 AM
OPM 1.0 does provide a number of performance graphs for a single volume (user workload). You can see the list here: https://library.netapp.com/ecmdocs/ECMP1364220/html/GUID-40A72B78-E550-4ACC-B823-C15162C7F0B8.html
2014-09-11 12:03 PM
I was wondering this same thing. If they are there, they're hidden very well. I found the OnCommand Performance Manager and installed that a few days ago, thinking that's where they were hidden...nope. While OPM does some cool things, it doesn't give you the info you're looking for often.
Sure, you can get to a volume IOPS/response time graph...if you search for the volume name in OPM. This is very unintuitive, there is no "browse" option - so you have to know the volume name and type it in the search box. Why can't you get to this from OCUM? Also, you can't see anything higher than a volume, so you can't get a stacked graph of "these are the busiest volumes by order of IOPS". Basically, its pretty useless for troubleshooting.
For example, one of my SQL cluster instances decided to suck a whole ton of IOPS due to a software issue, which was killing the storage system...causing B2B CPs and severely impacting other workloads. Without this per-volume IOPS chart, it would have been very difficult to narrow down. Luckily, this system is still on 7-Mode.
However, it appears that OCUM is substantially inferior to DFM/OC5...There are no real-time CPU charts, IOPS charts (per protocol or per volume), no throughput charts, no disk busy charts, etc.
Unless I'm doing something wrong...
2014-09-11 08:12 PM
looks like netapp wants to push out customers from system details, so they will always do a perfstat and contact netapp support ?? or is there another license or feature that we only have to pay for ?;(
and we surely have to pay for vmWare to run Monitoring and boxed analysis ??!
2014-09-11 10:12 PM
For the time being, it would probably be a good idea to install OCUM 5.x, in addition to OCUM 6.x, monitoring your CDOT systems. Performance Advisor remains available in 5.x for CDOT, and can provide some information that is not yet available in OPM.
2014-12-02 12:35 PM
OnCommand Performance Manager 1.1 shipped in November 2014. It includes a push API that allows customers to access the performance charts you reference in your post in the tool of their choice, such as open source charting tool Graphite.
2014-12-08 03:09 AM - edited 2014-12-08 04:04 AM
Correct me if i am wrong,
With clustered Ontap, DFM 5.2 is minimized in options. Now i can't create custom views anymore, so i can't add things like flash counters, or other "in depth" information.
Instead of DFM Clustered Ontap now uses Unified Manager, and if i need to see perfromane incidents i have to import another appliance calles Performance monitor.
All this results in less information in a more complicated storage enviroment?! And if i want more in depth information about my storage, i need to use the api and some third party tools to present it?
Or use powershell..?
2014-12-31 10:32 AM
Yossi hit the target in his reply above. For Clustered ONTAP since you have OCUM 6.2 you now definitely want OCPM 1.1 (still in release candidate stage I think, but worth getting it). While an exact equivalent of the Management Console from DFM isn't coming back, NetApp isn't cutting off users. OCPM 1.1 has the push feature mentioned that let you push statistics out to third party tools. It also has a MySQL database engine behind the scenes to which users can directly connect to get performance data from the system - much like the old "dfm perf" commands we had back in the day to get all the counters.
Like a lot of Clustered Data ONTAP, the performance tools have taken a while to catch up. I've run the dual setup (DFM 5.2/OCUM 6.X) for a while just so I could access the data collected for creating my own performance graphs, even though the data has been way more limited. Will be nice to drop down to one set of collection tools again.