FPolicy requests - how much is this?

[Sig]

Have a FAS9000 with SAS disk.  We have FPolicy enabled with an external-engine.

Right now we are doing approx 300-400 requests/second with spikes of up to 1000. 

I'd like to know if this is a lot?

How much is your system doing?

If anyone can run the following command and report back that would be awesome.

set d -confirmations off;statistics show-periodic -interval 1 -iterations 30 -summary true -object fpolicy -counter cifs_requests|aborted_requests|nfs_requests|processed_requests|requests_received_rate -instance VSERVERNAME

Thanks

Sig

[SpindleNinja]

I can't post output from customers, but from what I've seen across various customers,   300-400 isn't a whole lot of request,  especially for a F9K.  

Are you running in to issues or just curious? 

[Sig]

We have loaded about 30% of the expected workload.  Curious to know when we get to 100% (approx 1500 req/sec) if we will be at any ONTAP limit/maxes for FPolicy.  

If other customers are able to do 20k, 50k, 100k, etc then that gives a level of comfort that FAS9000 should be in a good position moving forward. 

thanks

[Ontapforrum]

I guess, it is difficult to predict, when the eventual workload (100%) is exercised, would it effect performance or not. But, I think there are few FPolicy counters (latency related) that would provide some idea. Also, if you start to see file-access latency/disconnections then it will definitely reflect in some way in the OCUM IOPS & CPU usage.  You may be able to see the errors/messages in the fplocy mlog if it is facing any pressure internally.

You can look for fpolicy_mlog messages:
/mroot/etc/log/mlog is location for fpolicy related messages.

Ensure you are running a fixed ontap version for this bug: (Increased FPolicy latency under a low or sporadic workload)
https://mysupport.netapp.com/site/bugs-online/product/ONTAP/BURT/1415152

Maximum outstanding requests on the FPolicy server: (Default-500, you might need to tweak this in future)
https://docs.netapp.com/us-en/ontap/nas-audit/plan-fpolicy-external-engine-config-concept.html#what-the-advanced-external-engine-options-are

Related kb:
https://www.netapp.com/blog/monitor-ontap-fpolicy-performance/

[SpindleNinja]

to add to what Ontapforrum said.  

Depending on your config - there's also the latency to the external server to contend with in your environment.  Or if you have sync or "async" type fpolicy configured as well.   

[Sig]

Thanks for the replies and links.  I understand that there are other factors when dealing with FPolicy performance. 

I'm just trying to get an understanding of the possible performance range for ONTAP.

IE: If I posted that my FAS9000 was doing 400 IOPS CIFS, folks could tell me that their FAS8300 was doing 20,000 IOPS CIFS.

400 FPolicy requests/second means nothing when I have no reference point.

Thanks

Sig

[Sig · Accepted solution]

Info on this has been very hard to come by so I'm going to close the loop with what info I have found.

Our current FPolicy numbers are 200-400 requests/s (spikes to 1000) (spikes ~3.7millons screen requests per hour) on a FAS9000 with SAS 1.8TB disks.
NetApp is unable to provide benchmark numbers/max limits for FPolicy.
An internet comment was that this is not a high workload.
An internet comment was that they have a NetApp system with "screen requests in the 20’s of millions per hour"

don't know if this helps anyone or not...time will tell 🙂