Virtual Disk Waste Calculation in vCOps

[TheVMinator]

If I provision a single virtual disk of 60GB on a virtual machine, install the OS and applications, and I have used only 30GB, how much virtual disk "waste" will vCOPs calculate? Is all of the 30GB that is not used yet considered "waste" or just a percentage of the unused space?  How is that calculation done?

Thanks

[mark_j]

In your scenario, the virtual disk space wouldn't be considered waste. vC Ops isn't going to look at the GuestOS disk space when determining waste.

Datastore waste is calculated by adding up templates and snapshot space.

[gradinka]

Adding to mark's reply, guestOS free space contributes to the TimeRemaining badge for a VM.
That is possible when you have vmware-tools installed/running isnde that VM.

[TheVMinator]

thanks for the information. I guess what I'm not clear about is if free space on the virtual disk is not used to calculate a virtual machine's virtual disk waste, what is? How does vCOPs make the calculation?

[gradinka]

here is some useful blog post:http://www.satisnet.co.uk/blog/vcops-5-efficiency-badge-reclaimable-waste-under-the-bonnet/

[TheVMinator]

the link represents a good thought process and helpful info for CPU waste, but I still don't see anything that would indicate what virtual disk waste in vCOps actually is.  I have a VM with an 80GB disk.  There is right now 20GB of data on it.

I go to operations -> All metrics and select a virtual machine.  Then I go to:

Waste -> Disk Space ->Total

How did that figure get calculated with what mathematical formula?

[gradinka]

1) In general waste score is a ratio of the amount of reclaimable capacity / amount of deployed capacity.

2) It is computed for cpu disk and memory (for each VM), and the largest value is used for the waste score on that VM

Percentage Reclaimable Capacity - it is the amount of capacity that is reclaimable, relative to the amount of capacity deployed (i.e. 20 vCPUs / 400 vCPUs).

• Powered Off VMs: VMs that can be deleted to recover disk space.
• Idle VMs: VMs that can be shut down and removed. This will recover CPU, Memory, and Disk Space.
• VMs with Old Snapshots: VMs that can trim disk space.
• Oversized VMs: VMs that can be shrunk to recover capacity

So to your question on the calculation, disk-space-waste is not about disk space _inside_ the VM, it is about disk-space which VM occupies on the host.

[TheVMinator]

Thank you for your response.  Can we narrow down the discussion by only focusing on disk, not memory or cpu. Thank you for itemizing the factors that compose the waste,I still don't see how the number is calculated based on the factors you discussed.

So I start with a VM with 80GB of disk.  The factors you mentioned are:

Factor 1: Has this VM been idle?  If so, what is the "idle" factor in calculating the formula such as:  if idle, that 80GB is waste, if not that 80GB is not waste?  How does a VM being idle affect the calculation disk waste?

Factor 2:  Does this VM have snapshots?  If so, how much snapshot size is considered "waste"?  The whole of all snapshots?  30% of the snapshots?  How does a vm having snapshots affect disk waste calculation?

Factor 3:  Oversized VMs - I have a VM with 30GB of actual data.  How much data is considered "oversized"?  For my 80GB vm, is 50GB considered the "over" amount?  I need some extra space to run the OS beyond the 30GB that it is used.  What does "oversized" mean in terms of disk space?

Also, can you clarify this statement:

So to your question on the calculation, disk-space-waste is not about disk space _inside_ the VM, it is about disk-space which VM occupies on the host.

This VM uses disk space on a LUN, it doesn't use local storage.  So it doesn't occupy disk space on an ESXi host.  So I'm not clear what you mean by this.

[gradinka]

Factor 3:  Oversized VMs - I have a VM with 30GB of actual data.  How much data is considered "oversized"?  For my 80GB vm, is 50GB considered the "over" amount?  I need some extra space to run the OS beyond the 30GB that it is used.  What does "oversized" mean in terms of disk space?

Also, can you clarify this statement:

So to your question on the calculation, disk-space-waste is not about disk space _inside_ the VM, it is about disk-space which VM occupies on the host.

This VM uses disk space on a LUN, it doesn't use local storage.  So it doesn't occupy disk space on an ESXi host.  So I'm not clear what you mean by this.

factor3->

the main thing is that disk-waste is about the disk-space which VM occupies as an object within your virtual infrastructure. That is the VMDK file(s), snapshots, logs, etc.

It has nothing to do with the situation inside that VM (e.g. you have 60GB free disk space on "C:" drive, how does that affect waste?. It does not!) .

I don't really know how to explain that in a different way

factor1, 2 ->

Those are configurable. Open vSphereUI->Configuration; Select 'Manage policies', pick a policy, do 'edit' and look at sections 3 and 4.

You define the rules when a VM is to be tagged as idle. But when it is, all the resources assigned to it are considered as a reclaimable waste. CPUs, memory assigned,  and disk-space used(total).

edit:

- Take a look at the "Datastore waste" view in vSphere->Planning->Views section. You will see the exact breakdown for a Datastore, contributing to disk waste. It is based on the VMs residing on it.

- the "Waste|Diskspace|Total" value -  if a VM is PoweredOff, this will show the total amount of disk-space occupied by this VM.  Snapshots and Templates are considered waste only AFTER certain number of days. See Configuration, 4c.

  What I'm not sure is whether snapshots contribute to VM waste or datastore waste. Have to check it when I have spare time. But it can be verified easily on test-instance with minor tweaking of the settings.

[TheVMinator]

OK thank you for the detailed reply. Let me try to clarify what I"m trying to do.  Suppose I am a VMware admin and I want to manage resources well.  Among my many goals here are some:

Goal 1:  I provision lots of VMs every day, and requesters always ask for bigger virtual disks than they need.  I don't want to provision 1000 VMs with 100 GB of vdisks per VM, when users are only going to ever use 30 GB per VM over the life of the VM.    At least from my perspective, if I did that it would be waste.  In other words, I want to manage the size of the virtual disks I provision responsibly.  I want to manage how large of virtual disks I create.  If users requested huge virtual disks for 100 VMs and a year later aren't using half the space they requested, I want to report on it. Do I understand correctly that you are saying vCOps can't do that?

Goal 2: I ask the storage team for LUNs every day.  In general I want to be able to view my datastores from the perspective that I am requesting the size I need, not too much or too little, and using the amount of space on that datastore responsibly.  I don't want to have too much unused space, and the space that has data on it is data that I actually need to have there (not powered off and unused VMs, etc). I want to manage space within my datastore well. This is what I would try to do by looking at waste from the perspective of a datastore. (That is a good goal, but it happens not to be what I'm after at the moment)

I go into the standard UI and select the "world" Object. I go to "operations->All Metrics->Virtual Machine->Waste->Disk Space to view info relevant to goal 1.  What you are saying is that this info is not intended to help goal 1.  So then where do I see info that would help me acheive goal 1 in vCOps?  I think goal 1 is pretty common in the industry so I would be very suprised to find that vCOps can't assist toward that.

factor1, 2 ->

Those are configurable. Open vSphereUI->Configuration; Select 'Manage policies', pick a policy, do 'edit' and look at sections 3 and 4.

You define the rules when a VM is to be tagged as idle. But when it is, all the resources assigned to it are considered as a reclaimable waste. CPUs, memory assigned,  and disk-space used(total).

So if I have a 100GB VM and it is taggled idle 30% of the time, how much of that 100GB is considered waste?  30GB?

It seems to me that somewhere or at some point VMware has to publish the details of what these specific places in vCOps are actually calculating and what is factored into them, if someone reading them is going to make practical use of them, so that the community has an objective reference point.

[gradinka]

So if I have a 100GB VM and it is taggled idle 30% of the time, how much of that 100GB is considered waste?  30GB?

-------------

you are mixing two concepts here.

there is

1) vcops configuration which determines when a VM will be tagged as idle. That is on the snapshot below

2) amount of disk-space (waste) which can be reclaimed by you for any such idle VM. That is not configurable.

For the purpose of it's calculations, vcops will either flag the VM as idle, or will not; There is no middle ground.
If it tags it, then you get reclaimable waste for that VM based on the fact that it's just sitting there doing nothing, occupying space.

So: If the VM is powered-off or tagged as Idle, all disk-space used by it is counted as waste.
Because you cannot reclaim only part of it - it's either all or nothing. e.g. you delete the VM or keep it.

[TheVMinator]

OK great thanks for the clarification.  So for the purposes of defining "reclaimable waste", am I correct in saying that there is no other factor used in determining whether a VM is considered reclaimable waste other than it's"idleness"?  Does that principle apply when defining reclaimable waste for all objects?  (I created a separate post )