Thin Provisioning

I get asked by a lot of admins about thin provisioning and how that data can be easily understood. Admins are looking to understand the risk in over provisioning datastores, because if you run out of space you can cause outages for several VMs that no longer can read and write to a disk. Unfortunately, there is no metric for a datastore that can show that, but this is where a derived metric can really help. With vFoglight, I can create a metric based on other metrics or anything really. So, in the case of datastores, I want to take all of the VMDKs, thin or thick, add up their committed/uncommitted space and create a datastore version of that. With this new set of metrics, as well as the existing VM level metrics, we can craft some really simple to understand, but powerful views.

Well, I've taken the liberty of creating a sample Thin Provisioning cartridge you can use to add these derived metrics to your environment and it also has several nice views around thin provisioning in general. You can check it out here:

Example views:

Drag out the View in Thin_Provisioning and it will give you a list of all VMs that currently have thin provisioned disks.

Drag out a datastore to get either the Graph & Table or just the Table. This shows the total allocation that is currently in use and data that could be written (think over provisioning)

Drag out all datastores to see all Datastores thin provisioning information