virtualization.info

continues from The Hidden Challenges of Virtualization - Part 2, published on virtualization.info on June 3, 2009

Data and Metrics of the Virtual Environment
The last entry left off with preparing a deck to sell a virtualization program.  The sales aspect will be further discussed separately in future entries to serve it justice.  Continuing with the data and metrics topic, this entry will focus on reporting the status and health of the virtual environments and infrastructure.

Assuming the virtualization program is now taking off like wild fire due to the save opportunities, there are a few things to beware of to keep the program running.  One is providing reports to prove the program is saving what the initial plan stated, and hopefully even more.  The key to doing this reporting is accurate inventory data and consistent processes for capturing virtual instances.  In order to properly track the data, the following elements should be captured:

• Virtual Infrastructure details (Hosts, Farm or Cluster info, etc.)
• Virtual Instance details (including in which Host, Farm or Cluster the instance resides)
• Is the Virtual Instance a net new request or a migration from physical to virtual
• If a migration from physical, what was the name of the physical server and what happen to the physical server after the migration (reused, sold, written off and donated, etc.)
  This data is needed to accurately track the growth of the virtual environment and calculate the saves related.  Weekly and monthly metrics should be prepared and sent to sponsors, etc. to show the growth and track trends related to saves from the program.

Other data related to storage and network use in the virtual environments compared to the physical environments can also be very useful and provide additional save metrics.  One very popular metric and save that virtualization is helping to drive is energy cost.  Power, cooling and data center space are natural saves for virtualization.  Although these saves are often not as large as the reduction in hardware from virtualization, the saves get a lot of attention based on the current environmental status of the world.  These saves can be hard to track, but here are some examples of how to track them and tie them back to higher level useful figures for sponsors.

Typically the virtual environments today are built in a farm or cluster scenario, especially in the x86 space.  This requires like if not identical hardware for the host servers.  Using the power and cooling calculators provided by the hardware vendors, calculate the watts used by the host servers and take the average physical server you are replacing and calculate the watts used for that server footprint as well.  The by using the metrics that track the host servers and new virtual instances a calculation for watts saved can be derived.  If there is an internal cost available per watt for power and cooling (many companies calculate this), then turn the watts saved by virtualization from either migrations or net new instance (cost avoidance) into a save and report it out to the key players.

In summary, data for metrics reporting it key to selling and continuing to fund a solid server optimization program. Next week the blog post will cover how to recover (rate cards, chargeback’s, etc.) for the program.