David Merrill

VM TCO Baseline Process

Blog Post created by David Merrill Employee on Sep 15, 2016

We get frequent requests from customers, partners and sales teams to do a VM TCO baseline. Over the years we have learned that an 'arms-length' cost analysis is an effective way to get started on a cost improvement program. This process is designed to be light-weight:

  • We use industry averaged and empirical data to estimate cost areas chosen by the client
  • Location, geographic rates are used whenever possible
  • Our cost estimation is usually 80-85% accurate, and tends to be good-enough to get an improvement plan started


The process to do this free assessment is as follows


1. We need to know what costs are important to the customer. This paper describes our VM costing methodology, and outlines 24 kinds of costs that can be used for a baseline measurement. Usually, people choose between 10-14 of these costs. Some of the costs may be hard costs, others soft cost; and we can discriminate between hard and soft in our calculation process.

Screen Shot 2016-09-14 at 3.27.51 PM.png

2. Provide the site location (city, state, country) where the assets are located. We use this information to determine local power, tax and labor rates


3. Next we need to have some details about the size and quantity of the VM systems and their hosts. The best way to get this data is to run a VMware script. The script is located here. Instructions for running the script are as follows:

  • Get both the Get-Info.ps1 and Get-RDM.ps1 scripts copy them to a machine that has VMware’s Power CLI installed
  • Run scripts once against each vCenter system that exists
    • Enter vCenter server name
    • Login with credentials that have a minimum of read-only access to the entire vCenter tree
    • Save output files Host-Info.csv, Datastore-Info.csv, VM-Info.csv, and RDM.txt into a separate folder for each vCenter System
    • Output files will be overwritten each run of the scripts
    • Repeat for each vCenter System
  • Zip the output files and send them to me at david.merrill@hds.com


4. If for some reason the scripts cannot be run, then this form can be used to manually input the data that we need.


That is about it. Within a few days, HDS can create a model and a corresponding report that describes:

  • Total average cost per VM
  • Average cost, by VM size
  • Observations on the costs
  • Ideas to reduce unit costs


With a starting point on VM unit costs, IT managers are in a position to start adding precision to the models, and then make plans on how to materially improve VM infrastructure, operations and VM consumption behavior in order to drive down unit costs.