Silicon Valley technology companies and California businesses fear they are losing competitive edge and being priced out of the market by power supply disruptions and increasing energy costs according to delegates attending DatacenterDynamics, 27th July at the Grand Hyatt, San Francisco. In the week that the California Independent System Operator declared a Stage 1 Electrical Emergency, imposing rolling brownouts to counter high temperatures and power demand, conference delegates discussed with industry experts, the need to drive power efficiency at every level of the data center. In fact, several delegates were unable to attend the event this year as they stayed onsite wth their facilities monitoring thermal loads and the operations of their backup generators.
Stephen Worn, conference chairman said: "A typical data center uses as much as 50 to 100 times more electricity than a comparable size office building. If you set the growing hunger for power against the backdrop of a region whose electricity reserves are being pushed to limit, then it's not hard to see why power is such a big issue in California and specifically the Bay Area and Silicon Valley. California's Energy Commission states that the region's data centers have a combined energy requirement totalling an estimated 375MW. The dilemma is how to support a growing demand for data center services whilst reducing the overall power consumption.
He continued: "Unsurprisingly, given the events of the week, electricity was the hot topic on the lips of delegates. Many attendees complained of spiraling electricity costs which have set operational expenses soaring and taken a sledgehammer to profits. From performance per watt at chip level through to on-site generation opportunities, and new solutions in rack level thermal management, delegates at the show were keen to seek out technologies and techniques that would help reduce power consumption and increase electrical efficiency.
Now in its third year, the San Francisco event attracted a record 427 delegates from many of California's leading data center operators. DatacenterDynamics events provide a knowledge exchange for professionals involved in the design, construction and management of datacenters. The conference is part of series of 14 educational and networking conferences being held at major datacenter locations around the world aimed at providing a global perspective of the datacenter industry.
Delegates gathered to hear keynotes and seminars delivered by acknowledged data center experts including Andrew Fanara of the Environmental Protection Agency (EPA), Energy Star Program and Dr. John Koomy, Lawrence Berkeley National Laboratory (LBNL) who jointly outlined their road map to help US industry reduce power consumption. The presentation reported on the progress of joint server energy efficiency research and server performance benchmarking metrics and outlined how the US government is working with manufacturers, operators and leading institutions to encourage efficiency and seek solutions.
Also discussed was the Data Center Efficiency Bill (commonly referred to as H.R. 5646) which has been approved by the U.S. House of Representatives and is awaiting Senate and Presidential approval, as well as calls for a six-month Environmental Protection Agency (EPA) study on data center efficiency.
Green issues were also on the agenda as David Winn and William Kosik of EYP Mission Critical Facilities showcased one of the world's first energy efficient green' datacenters. In their presentation Going Green' they demonstrated how in one case, onsite critical energy systems are utilising gas turbines to reduce energy costs by more than 30% and green-house emissions by 70%.
Other DataCenterDynamics highlights included:
Maximising Cluster Computing while Minimising Power Consumption and Heat Generation' - Brent Kerby of AMD and Jeff Stilwell of Rackable Systems outlined the elements that contribute to keeping data center costs down including low platform power consumption, leveraging existing power infrastructure, advanced service oriented architectures, chip and server level energy and cooling, as well as utilizing space efficiently.
Building a data center with a 20 year lifespan' - Carrie Higbie from Siemon exposed the myths of blade servers and presented real world examples of issues, design, operational total costs of ownership and technological factors to the delegates. This session investigated fiber infrastructure and a future roadmap for IT and facilities managers.
The conference highlighted a number of issues for the community over the next 6-12 months:
Ôé¼. Tier classification - The Uptime Institute's Tier Performance Standards have long been the de facto measure for facility resilience, used as an objective basis for comparing the capabilities of a particular design topology against others or to compare groups of sites. The conference highlighted the need to expand on tier systems for facilities. The session by Syska Hennessy, Revisiting the data center classification model' demonstrated the industry need to discuss this issue and come up with a modern and more robust classification system.
Ôé¼. IT and Facilities Management integration - As the roles of IT and FM management are increasingly integrated in the datacenter, the conference highlighted the requirement for proactive management and automation tools and systems which bring together co-owned functions such as power, cooling and security.
George Rockett, conference director noted: "Once again DataCenter Dynamics San Francisco has proved a hot-bed for education and information exchange, as well as providing a positive sounding board for the discussion of technologies of the future. Power consumption undoubtedly dominated proceedings but delegates also showed an appetite for vendor innovation and solutions. For example, as the demand on racks is changing and IT budgets are squeezed, many datacenter operators do not have the funding available to replace rack infrastructure. One delegate asked when will the transforming cabinet be available?' If his kid could have a toy that could transform from an automobile to a robot, why couldn't the finest brains in infrastructure design create a cabinet that could also transform according to prevailing datacenter power and thermal rack level requirements?