The Frontier Grid Platform
Server Consolidation & Virtualization
Server Consolidation through Utilization and Virtualization
The availability of low-cost servers over the past decade has fostered a pervasive tendency among organizations to purchase computing capacity far in excess of actual demand. Studies show that aggregate server daytime utilization varies between 5-20%1. Some experts believe aggregate utilization rates fail to exceed 5% when desktops are included.2
Overcoming the Power, Space & Cooling Problem
The Frontier Enterprise Grid Platform is a great tool for reducing the power, space and cooling issues that accompany unchecked "server proliferation." By harnessing the excess capacity of existing infrastructure, Frontier can make available the 80-95% of an organization's unused computational capacity — without impacting normal operations! When HPC demand is satisfied in this way, fewer new hardware resources need to be procured and maintained, which greatly aids server consolidation efforts.
A Frontier Virtual Appliance
Additionally, the Frontier Grid Platform is available as a set of Virtual Appliances (VA), allowing easy deployment and management in virtualized environments. The Frontier Compute Engine VA can be configured to run at set times, use only a portion of available computational power, and give priority to other processes and VMs. A single Frontier Server VA can manage a grid of thousands of nodes and because it is horizontally scalable, additional Frontier Server VAs can be deployed side-by-side to dynamically accommodate grids of tens of thousands of nodes. Without affecting other applications in the data center, excess capacity can be turned into supercomputing capacity.
By exposing the full capacity of an organization's computational resources, physical or virtualized, Frontier Enterprise provides HPC capacity while simultaneously reducing power, space and cooling demands.
1 Taurus - A Taxonomy of the Actual Utilization of Real UNIX and Windows Servers, David G Heap, Principal IT Consultant, IBM Enterprise Server Group, January 2003.
2 Grid Computing in Financial Markets: Moving Beyond Compute-Intensive Applications, Larry Tabb, Tabb Group, November 2003. http://www.tabbgroup.com/PublicationDetail.aspx?PublicationID=68