As virtual environments gets more complicated and challenging to manage, we have to figure out how can we do more (work…) with less (time…).
In today’s virtual environments – it’s quit common to find VM’s with many vmdk’s attached to it, and it can be from many reasons – capacity, performance, separation of OS disk and application disks, etc…
One of those cases that can be very time consuming is to locate that single “monster” vmdk that doing a lot of IOPS.
Just try to imagine how long it will take to find that one heavy utilized vmdk among 10, 20 and even 30 vmdk that the virtual machine is using.
vCOps custom dashboard to the rescue!
Using vCOps custom dashboard – we can do it with no sweat!
So – lets begin!
1. Creating the custom dashboard
Under “Dashboards” , click on “Add”:
For this custom dashboard, we will use the “Top-N Analysis”, “Metric Selector” and “Metric Sparkline” widgets.
Just drag and drop the widgets from left to right.
2. Configure “Top-N Analysis” Widget
Next, we will want to find out who are the top-10/15/20 (choose you favorit flavour) IOPS VM’s.
For this dashboard – the IOPS data will be based on the last 1 hour.
Self Provider – Off
Image redrew rate – 5 minutes (in other words – refresh rate)
Period Length – Current Value
Widget Mode – Metric
and then World -> Virtual Machine -> Virtual Disks -> aggregate all instances -> Commands per seconds
3. Creating custom XML for VMDK mapping
your probably asking yourself – “how can we display each VM’s vmdk, since every vm can have different number of it”.
Well – fair question!
There isn’t any sophisticated or elegant way to do it beside mapping each 64 possibilities (4 SCSI Controllers x 16 vmdk per SCSI controller).
Here is the example of the SCSI 0 controller:
and same goes for each and every additional controller (thanks god for CTRL+C, CTRL+V)
Using WinSCP – copy the XML file to the following path:
5. Configure “Metric Sparkline” widget
Since we are using custom XML, there isn’t much configuration needed besides:
Self Provider – Off
Res. Interaction Mode – Select the custom XML
6. Configure “Metric Graph” Widget
Very similar to “Metric Sparkline” configuration
7. Configure Widgets interraction
We want to configure interaction in order that for any listed VM in the “Top-N Analysis” – we will be able to get the data on both Metric Sparkline and Metric Graph widgets.
Go to “Interaaction” and configure the following
8. That’s IT!
There you have it – Top IOPS VM’s per VMDK!
wasn’t so hard, was it? (:
As i’m writing this lines, i just thought on something extra that can be added to this dashboard:
When finding which one of the vmdk is doing most of IOPS, we can then use the Resource widget to discover on which datastore it relays on!
I’ll leave this small tip for you to do (:
vCops Custom Dashboard – So What did we have here?
With a click of a button – now you can know not only who are your top IOPS consuming VM’s, but also how IOPS are spread over the VM’s VMDK.
Is that awesome or what?
That’s all for now folks, but don’t worry – more good stuff are ahead.