Virtualization Capacity Planning

No other technology in the last 5 years has had a bigger impact on reducing power consumption in the data center than virtualization.  Yet, many companies aren’t seeing significant reductions in power usage and the reduce PUE’s that should be a natural consequence of implementing this technology.  Part of the problem can be traced back to the beginning of these projects and the leap of faith that data center managers took based on the claims of the virtualization vendors.  That virtualizing applications will reduce energy consumption in the data center regardless of the implementation process.

A recent survey of 300 data center managers conducted by Quocirca Research about the level of server utilization in their facilities provides some interesting insight.  Of the 300 data center managers surveyed, 35% estimated that their servers were utilized over 50% of the time and a significantly higher percentage claimed a higher utilization percentage than the widely accepted 5% – 10% utilization figure for the industry.  Our conclusion from these findings is that data center managers aren’t accurately measuring server utilization and typically don’t baseline their metrics before starting a virtualization project, otherwise we wouldn’t see such a wide gap between perception and reality.

The real gains in power reduction based on virtualization technologies is found in the decommissioning of older, less energy efficient servers that are consuming power and not doing useful work.  The nlyte Data Center Infrastructure Management (DCIM) suite provides a systematic series of process steps that helps data center managers better plan and implement successful virtualization projects.

With nlyte’s modeling capabilities and real-time power collection, you have an accurate baseline of your facility’s power consumption right from the start.  The nlyte Discover module will accurately collect server utilization metrics and the applications on these servers so you can model the cascading affects of changes in this environment.  You can identify a certain class of server, model the affects of virtualizing all the applications supported by these servers and measure the affect of decommissioning this equipment on the facility’s power capacity.  With nlyte’s out of the box integration with VMWare™, the virtual hosts are mapped directly to the physical servers in real-time, so an accurate depiction of your data center’s virtualization capacity can be derived.  This virtual host mapping reduces the risk of creating “virtual” hotspots and protects against “virtual” sprawl.

When your modeling efforts have produced the desired results, nlyte’s Control Module will help direct your personnel to implement these changes in a orderly and controlled fashion according to ITIL® and COBIT best practices.