Virtualization Technology News and Information
Xsigo Systems Offers Seven Tips to Leverage Virtual I/O for Greener Data Centers
As companies realize that greening the data center is an ideal way to reduce electricity use and minimize environmental impact, they are quickly embracing server virtualization and VM machines to help reach this goal. The growth of virtualization, however, gives rise to a new problem -- server connectivity and I/O infrastructure bottlenecks. Xsigo Systems, Inc., the technology leader in data center I/O virtualization, today issued a list of seven tips to help IT communities leverage virtualization, while removing known risks and barriers.

"As more servers become virtualized, it only makes sense to virtualize the supporting I/O infrastructure," said Vinod Khosla, founder of Khosla Ventures and investor in companies that contribute to green computing, including Xsigo Systems. "Newer virtualization technologies such as virtual I/O, combined with 'green' best practices and management, will absolutely allow data center energy consumption growth to slow over the next few years."

To help IT managers leverage I/O virtualization for greener data centers, Xsigo published the following tips.

1) Eliminate the anchor points that limit flexibility

Virtual machine technology delivers part of the solution needed to increase data center efficiency, but make sure to provide sufficient network and storage connectivity to allow maximum flexibility on virtual machine deployments. Virtual I/O helps by eliminating the fixed connectivity that binds applications to specific servers and limits the ability to optimize resource utilization.

2) Banish the roadblocks to virtualization

To enhance virtualization adoption, eliminate blocking issues such as security concerns. Plan for I/O that can provide each virtual machine with secure connections to the needed storage and networks. Virtual I/O simplifies this by providing multiple connections on a single cable. This ensures security for multiple applications per server without creating cable bloat.

3) Put your servers on an I/O card diet

In an I/O intensive server, network and storage connectivity may consume up to 50 watts. Factoring in the power required to cool the system, this doubles to 100 watts. Virtual I/O eliminates 70% of the I/O cards, bringing consumption down to 20 watts per server. In a data center with 1500 servers, this translates to an annual savings of 1 megawatt hour, or enough electricity to power more than 100 typical homes.

4) Combine virtual I/O and storage virtualization to maximize savings

Newer storage virtualization technologies such as thin provisioning let you allocate capacity without actually having spinning disk consuming power. When storage requirements grow, disks can be added without disrupting the application. Similarly, I/O virtualization lets you connect storage to a specific virtual machine. When processing requirements grow, the virtual I/O and the virtual machine can be moved to another server, again without disrupting the application. These technologies provide a seamless path to the future without wasting power in the present.

5) Get high availability on low power

High availability becomes simple when virtual machines are portable among servers, making virtualization a cost effective way to boost uptime. But to achieve full virtual machine portability, the primary and failover servers must have the same network connectivity. Virtual I/O provides the needed connectivity to every server without running a dozen or more cables to each. The result is fewer cables, less hardware, and less power.

6) Address your constituencies

Build user confidence by implementing an infrastructure that will ensure application performance and availability in the virtualized world. Traditional I/O limits connectivity, which results in bottlenecks and increases the cost of redundancy. Virtual I/O maximizes the use of each physical connection to ensure maximum throughput and resilience, so users see faster performance and better availability.

7) Think globally, act globally

The EPA has set a goal of 5 to 1 overall server consolidation to stem the growth of power consumption. Achieving this goal will require fundamental changes. Servers must become a far more interchangeable resource, with the IT managers controlling server, networking, and storage resources acting as partners allied to boost efficiency. This may blur some management boundaries as the underlying resources are virtualized and transitioned to a new, more flexible control model. But by promoting streamlined coordination as an objective, these managers achieve the triple benefit of creating a more responsive data center, lowering overall costs, and, of course, helping the planet.

Published Tuesday, October 23, 2007 9:00 AM by David Marshall
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<October 2007>