VROPs monitor IOP per vmdk

[ICEMAN47]

We have allot of split VMs some is ssd SAN and some in spinning. I was wondering if vrops could monitor the read and write iops for each vmdk.

[carvaled]

Yes the metric will be per SCSI I'D under VirtualDisk

[iceman47777]

Trying to create a Dashboard that would show each Drive. looks like it cant do it for individual vmdks. only the sum of all.

[sxnxr]

be leave it or not but this worked allot better in 5.x than in 6.x

The way to create it is the same but the output is different.

Create a custom interaction file with the below (keep adding the metric key so you include all the possible SCSI IDs

<?xml version="1.0" encoding="UTF-8"?>

<AdapterKinds>

    <AdapterKind adapterKindKey="VMWARE">

        <ResourceKind resourceKindKey="VirtualMachine">

            <Metric attrkey="virtualDisk:scsi0:0|commandsAveraged_average" label="" unit="" yellow="" orange="" red=""/>

            <Metric attrkey="virtualDisk:scsi0:1|commandsAveraged_average" label="" unit="" yellow="" orange="" red=""/>

            <Metric attrkey="virtualDisk:scsi0:2|commandsAveraged_average" label="" unit="" yellow="" orange="" red=""/>

        </ResourceKind>

    </AdapterKind>

</AdapterKinds>

Then create a new dashboard with an object list focused on the VMs you want. Add in a metric chart and configure it with self provider off and use the custom interaction file. Set the widget interactions so the object list feeds into the metric chart.

Now this is where it is different. In 5.x when you selected a VM and it didnt have the disk for SCSI0:3 it would display id 1 and 2 bur show no graph for 3 but now in 6.x (have tested in all 6.x up to 6.6) if the disk is not present then it will show a graph with no data.

I used to love this dashboard because i had 170 lines in the metric configuration showing all the disks read, write and total IO per disk now it will show me 3 graphs with data and 167 graphs with no data.

This was the best way i could think of to do what you wanted

[ICEMAN47]

Thanks for responding, it's so frustrating that it would show a bunch of fake graphs. I have a mix of server with 1 to 9 drives.  Think we might end up going into solarwinds. Honestly would expect more from vmware but I guess it's going downhill.

Thanks again.

[sxnxr]

I would not say it is going down hill. Yes there are challenges and things vrops does not do but it is constantly improving. I have monthly calls with the vrops engineering, development and product team managers to discuses the problems/challenges we have. For example i brought up

1. No report for VM to datastore relationship
2. No in guest memory stats unless you use the EOPS agent which messed up the dashboarding due to two different adapters
3. No vmware tools versions collected

All these have now been added into the product. I think they are now realising they need to design the product based on what the customer needs which they are doing more of but yep it is not perfect.