Akamai Technologies – Shruthi Rao – 2010

Goals

Akamai Technologies enlisted Shruthi Rao, an MBA from Thunderbird School of Global Management, as its 2010 EDF Climate Corps fellow. Akamai provides cloud-based services for optimizing Web and mobile content and applications, online HD video and secure e-commerce. Akamai’s global platform, consisting of a network of over 70,000 servers in more than one thousand co-location data centers, is a major operational cost and source of carbon emissions. Fixed-power contracts at these co-location facilities provide Akamai no financial incentive to reduce off-peak power consumption of its servers. Rao’s challenge was to overcome this barrier to greater energy efficiency.

Solutions

Rao pursued a three-pronged approach: (a) analyzing the trade-offs between co-location vendor fixed and metered power contracts, (b) analyzing the energy efficiency potential of implementing server power management and (c) assessing potential cost savings to build a compelling business case for implementing server power management.

Rao strategized on how Akamai should proceed by comparing existing contracts with alternatives offered by the vendors, and estimating cost savings if it switched to metered power.

Potential Impact

By applying identified off-peak server efficiencies and using a metered model on a sample datacenter, Rao calculated that Akamai could reduce its electricity consumption by approximately 20 percent, translating to similar cost savings and a reduction in carbon emissions by approximately 1,200 metric tons over a three year time span.

In addition, she recommended adding sustainability requirements to vendor RFPs, such as requiring them to have certain PUE levels based on seasonal and geographical variations, which could also result in significant cost savings and reductions in carbon emissions.

If this model is applied to its most cost-intensive co-location facilities, Akamai could save millions of dollars and reduce its carbon footprint substantially over the same three year time span.