« Back to Home Page

Principles and Practices of Enterprise IT Efficiency

Microsoft | Tuesday August 30th, 2011 | 1 Comment

We partnered with Microsoft to run key findings from a recent white paper entitled “The IT Energy Efficiency Imperative” as a ten-part series. To read the full series, click here. The white paper can be downloaded here in PDF format.

Taking into account all of the opportunities for increasing IT energy efficiency, how can an organization best approach the challenge of making its IT operations more sustainable? Here are the key principles for a successful transition to operating IT as a utility:

  • Improving IT energy efficiency will significantly reduce the financial and environmental costs of running IT and make IT more responsive to business demands.
  • Increasing IT resource utilization is the most effective way to improve IT energy efficiency.
  • Central control of the IT infrastructure (local or cloud-based) is a key requirement for increasing IT resource utilization.
  • Applications designed to optimize IT resource utilization are necessary to maximize the effectiveness of a centrally controlled IT infrastructure.

The following practices support the aforementioned principles:

  • Continuously measure, report, and set goals for IT resource utilization.
  • Ensure that the costs of deploying and operating applications on dedicated hardware are fully reflected in the cost charged to the application owner—including energy use, allocated power, and data center space.
  • Centralize the budget for server hardware capital expenditures, and decentralize the budget for server hardware operations.
  • Use centralized computation-and-storage services (based in private and/or public clouds), and rent out capacity to business units to recover costs.
  • Adopt application designs that enable dynamic resource allocation, cloud bursting, controlled deferral of noncritical work, and workload power efficiency.
  • Implement power management on the desktop, ensure that applications function correctly with power management enabled, and enact policies that mandate the procurement of energy-efficient hardware.


Setting the Right Goals

Energy efficiency changes in one layer of the IT stack can affect energy efficiency in other layers. For instance, improving the energy efficiency of servers in a data center by 50% can also save energy at the building layer because less energy is required for power distribution and cooling. If you reduce server power consumption and you don’t add new servers to consume the power difference, you could end up actually making the PUE worse because the cooling systems in many data centers are sized to extract a specific range of heat. This might seem like a bad thing, but if you are producing at least the same amount of useful work with less energy, it’s worth the sacrifice. This is one examples where understanding the metrics is crucial in achieving one’s goals.

Many of the enabling technologies needed to implement these changes are already available, and the rest are on many IT vendors’ roadmaps. Perhaps the greatest barrier to organizations adopting this vision is inertia.

Fortunately, there are many opportunities to improve energy efficiency at all layers of the IT stack without re-engineering the entire IT operation.. The layers in the IT stack are interdependent, and their boundaries are not entirely distinct. For instance, operating system power management can be a coordinated effort between the Silicon and Operating System layers and can even include the Hardware Package layer, using firmware that enables cooperative processor power management. For an organization with multiple data centers, the Management Infrastructure layer will likely extend into the Building layer to allow for management of applications across multiple data centers.

While it is important to keep the end goal in mind—operating IT as a utility—IT energy efficiency can be improved significantly through the use of components that require less power to run, by managing components so they do not need to run continuously at the same power levels, and by reducing the amount of hardware needed to do the job

▼▼▼      1 Comment     ▼▼▼

Newsletter Signup