IT consolidation

Research opportunities

Growing businesses, government agencies and other large organizations are adding hundreds or even thousands of servers annually to meet ever-greater demands for speed, power and functionality. Yet many of these same enterprises are not making full use of the servers they already have.

According to a 2005 study by Forrester Research Inc., companies typically run their Windows servers at only about 8 to 12 percent of capacity, while UNIX servers usually operate at 25 to 30 percent of capacity. Meanwhile, both new and existing servers -- possibly from multiple vendors -- take up space, run up energy bills and require constant maintenance.

The solution seems obvious: Determine each server’s current workload and total capacity. Then consolidate information onto fewer servers, tapping more of each machine’s capabilities and perhaps using virtualization technology to make the reconfigured infrastructure even more flexible and efficient. But in today’s increasingly complex IT environments, that’s easier said than done -- and next to impossible to accomplish manually.

In fact, the typical large enterprise lacks a good big-picture view of its server population. Businesses also typically don’t fully understand the relationships between all those servers. For instance, they may not realize that an application on one server depends upon being able to reach a database stored on another -- which might be located on a different tier.

Research focus

Current work

We’re currently using or investigating methods for:

• analyzing IT environments and recommending options for consolidating workloads onto fewer servers based on customer objectives and constraints
• integrating server relationships and dependencies into the consolidation algorithm to minimize disruptions to the environment and ensure successful migrations
• leveraging storage traffic data captured through the server workloads to recommend SAN designs
• predicting application performance based on the underlying infrastructure. Such prediction is critical to IT organizations and service providers to meet user requirements and service-level agreements (SLAs), and can be used to identify optimal platforms and configurations

Technical contributions

HP is already making server consolidation simpler and more accurate with the Consolidation Analysis Tool (CAT), which is now being used in customer engagements. Based on Microsoft’s .NET architecture, the tool was developed with expertise throughout HP, including the Consulting & Integration group and HP Labs.

CAT assesses a customer’s IT environment, analyzing server inventory and performance data such as CPU and memory usage, over multiple days or weeks. The tool is designed to automatically deal with issues such as missing data or servers that should be excluded from analysis because they’re performing abnormally. Then it generates customized recommendations for server consolidation based both on that comprehensive analysis and on the customer’s specific objectives.

HP Labs developed the optimization algorithm -- the mathematical model that, combined with HP Services expertise, identifies the most promising opportunities for consolidation. The resulting tool is transforming the server-consolidation process from one based on best guesses to one driven by comprehensive, reliable, up-to-the-minute data -- and the customer’s ultimate needs and goals.

Learn more