Data centers have had the same problem almost from the beginning. How do data center operators justify their existence (and budget) when their whole goal is to do their job so well that executives and end users never notice they exist?
Statistics and reports about uptime, lines of code written, servers utilized or applications integrated have some impact, but don't communicate the amount of effort and expense that goes into making sure nothing happens.
One company, eBay, has combined the statistical approach with a tool, Digital Service Efficiency, that turns IT performance statistics into a visual element that a CFO can chew on: a number that rates the efficiency of a data center according to the actual flow of the business that IT is supposed to facilitate.
Digital Service Efficiency appears as a dashboard that displays power load, number of users, energy supplied and other measures of IT effectiveness along with the amount of revenue generated per user, cost of employing that user, carbon footprint and other measures of business activity.
eBay's Digital Service Efficiency dashboard visualizes key operational metrics for its data centers. Source: eBay
DSE is "the miles-per-gallon measure for technical infrastructure for eBay," says Dean Nelson, eBay's VP of data center strategy, as quoted in an eBay video announcing the service and other publicity materials.
The online auction site has been working on DSE and its prologues for about 18 months as part of an effort to become more efficient. The goal is a 10% increase in the number of financial transactions it can log per kilowatt-hour of electricity. It also wants 10% reductions in its cost per transaction and carbon use per transaction during 2013, compared to a baseline the company established in 2012.
The intent is to make eBay more efficient and profitable, but also to increase the efficiency of its power use in ways that contribute to the operation of the company, not by simply reducing the number of kilowatt hours of electricity it consumes, according to a brief describing the project that eBay posted for download.
In one case eBay said it saved about $2 million in capital costs by reducing the amount of memory used by a single application across a pool of servers. Reducing the memory requirement eventually allowed it to eliminate 400 of the servers, according to eBay's brief.
Inefficiencies highlighted by DSE also allowed eBay to identify layers of redundancy and fault tolerance that were, themselves, redundant. Eliminating them helped double the power capacity of its data centers, increase the density of its server racks four-fold and cut the cost-per-megawatt of power used in the data center by 53%.
During 2010, data centers used between 1.1% and 1.5% of all the electricity consumed globally, according to a 2011 study by Jonathan Koomey, a consulting professor at Stanford University and former director of the Lawrence Berkeley National Laboratory's End-Use Forecasting group, which studies the economic and environmental efficiency of new products and technologies.
[ Join us at Interop Las Vegas for access to 125+ IT sessions and 300+ exhibiting companies. Register today! ]
Koomey has also been one of the leading voices on the efficient use of electricity in IT with research including his 2007 analysis of the amount of power wasted by physical servers--which contributed to the cost-justification for server virtualization in corporate environments. He also published reports on electricity use by data centers that showed massive waste caused by poor HVAC design and usage policies.
The amount of electricity used by data centers worldwide doubled between 2000 and 2005, his studies showed. By 2010 that growth had slowed to the point that usage increased only 56% between 2005 and 2010 worldwide, and only 36% in the United States.
Saving power is only one intent for eBay's DSE project, according to the company. By correlating energy use with specific activities in IT and on the business side, DSE helps identify processes or technologies that operate with particular inefficiency, or which can be tuned to improve bottom-line results.
The calculations behind the DSE are designed to use and expand on the Power Usage Effectiveness (PUE) metric created by The Green Group, an industry consortium created to develop measurements and technical standards to make IT hardware more energy efficient.
The code behind DSE is too company-specific to be useful to other data center operators, but eBay will release materials describing its methods in gathering data and calculating metrics as a way to encourage other companies to follow its example, according to the DSE brief.