Active IQ Unified Manager Discussions

Unified Manager metrics vs grafana and netapp harvest

zenmatrix83

We've only had grafana setup with harvest 1.4 for a little while and I'm noticing a difference the iops for volume seems way off compared to ocum. For example I have one volume thats showing anaverage of 7.7k iops over the last 24 ours, but ocum sees an average of 2.5k iops over the last 24 hours. I'm trying to figure out which one is correct, I would think ocum, but I have a qos policy on this volume and it seems to be pretty close to the 8k max iops policy I have set.

 

 

1 ACCEPTED SOLUTION

yannb

@vachagan_gratian Thoughts ?

 

I would zoom in and compare a small time sample, it might be explained by the granularity of Harvest campared to OCUM, which only works on a 5 minutes sample, but even then it shouldn't be that different.

 

Also check the type of IOPS considered in OCUM. I remember sometimes it didn't show other_ops, but I checked now on a 7.2 version and it actually does.

View solution in original post

9 REPLIES 9

zenmatrix83

I'm wondering now if we have something wrong, I saw a large workload just start. In the top average iops show as almost 20k in grafana, but in ocum and vmware they only show around 2k.

 

 

zenmatrix83

Never mind ocum in another screen shows the 20k, now I just need to figure out why my qos policies aren't working.

 

2019-07-08 14_19_26-.png

Thanks for the update and screens. Let us know if you need further help.

yannb

@vachagan_gratian Thoughts ?

 

I would zoom in and compare a small time sample, it might be explained by the granularity of Harvest campared to OCUM, which only works on a 5 minutes sample, but even then it shouldn't be that different.

 

Also check the type of IOPS considered in OCUM. I remember sometimes it didn't show other_ops, but I checked now on a 7.2 version and it actually does.

View solution in original post

vachagan_gratian

OCUM documentation says that those performance values are averaged over the last 72 hours, might that be the reason of the difference? (I couldn't find out if they include other_ops in the total value, but that's a good point Yan!).

 

In either case, Harvest gives us the exact values that are reported by the CM of the ONTAP system.

Thanks, guys, that's where I was leaning to from what I've read, it just seems to be a bit much since the same 24-hour graphs in both systems look relatively flat so I thought it was odd.

The difference is indeed too large to be neglected. Could you maybe send me a sceenshot of both dashboards? I'll try to investigate this when I find a bit of time.

ocum24iops.pngharvest24iops.png

 

 

zenmatrix83

Thanks, even the drawn out lines looks similar they are just about 5k off.

Announcements
NetApp on Discord Image

We're on Discord, are you?

Live Chat, Watch Parties, and More!

Explore Banner

Meet Explore, NetApp’s digital sales platform

Engage digitally throughout the sales process, from product discovery to configuration, and handle all your post-purchase needs.

NetApp Insights to Action
I2A Banner
Public