Data center dynamics

Power consumption and cooling of mission critical IT facilities remains a key concern, according to professionals who attended DatacenterDynamics New York.

For TFM‘s coverage of this topic, see “Ensuring The Data Center” from the archives.

George Rockett, conference director, DatacenterDynamics says, “The data center industry may be booming, but there is a fundamental need to address power, cooling, and knowledge issues. Increasing power densities, whether they are from networked storage, communications equipment and or more commonly from servers, are all contributing. The move to towards blade servers for example, has enabled dramatic increases in server density and performance with improved flexibility, reliability and reduced costs. However, the associated power and cooling requirements are necessitating a redesign of existing data centers as well as a rethink of the way that new data centers and LAN environments are deployed. Although the introduction of multi-core processors is touted as one solution, they are a short term answer to what is a long term problem. From ASHRAE to IBM and APC the delegates discussed the design, architecture and implementation methods for dealing with blade server cooling, and options and practical solutions for new, retrofit and existing environments.”

Rockett continues, “Another key issue facing large organizations is whether ownership of the data center should reside with Facility or IT management. There exists a lack of understanding between these two disciplines which is undoubtedly leading to problems. IT wants to run more applications and demands greater scalability, and this requires better hardware. But adding more hardware presents engineering problems and IT needs to recognize this. Similarly, the guy providing services into the datacenter needs to understand that as value of data has grown, the high cost of downtime and the penalties served by regulatory bodies have brought facilities sharply into focus for corporate executives.”

He concludes “What’s clear is that these are truly global problems. We have delegates attending not just from local markets but flying in from all over the globe to attend our events and they are all talking power, cooling and the need for standardized training.”

Other topics discussed included:
The return of liquid cooling to data centers. A key theme was the return of chilled water to data center environments. Stephen Worn, global conference chairman says: “Given the shear power needed, the requirement for liquid cooling to remove heat generated from IT equipment, isn’t just an option, in many cases it is the only solution. As the IT market continues to focus on maximizing the number of server blades per enclosure, traditional air-cooled server cabinets often cannot support the dramatic escalation in heat loads and new solutions are required. Enclosure-based liquid-cooling solutions not only represent next-generation cooling strategies, offering more effective heat transfer and removal from high-density installations, they are the ideal solution for the compact environments within the New York metropolitan area.”

“Conventionally cooled data centers are facing almost overwhelming problems regarding the cooling of high performance servers, with CRAC systems having a notional cooling limit of 3-5kW per cabinet. The problem many delegates face is that their cooling requirement per rack greatly exceeds this level. The event showcased very latest water-cooled thermal solutions which offer a dramatic advance in cooling capacity and can also achieve significant reductions in overall operating cost. We saw a range of solutions discussed and insight into the latest cooling options presented by experts from around the world, solutions included everything from entry level up to 35kW per single cabinet.”

Energy efficiency. The need control energy bills was also a hot topic, Worn comments, “As loads increase and vendors continue to pack more servers into a smaller footprint, energy costs rise as does the demand on the mission critical infrastructure. The cost of power and its efficient delivery was one of the top issues throughout the day, and discussing the ‘The Power Gap’ came from every direction – from maximizing data center energy efficiency through power-saving designs, technologies and DC-based solutions, through to scalable UPS systems, and battery versus rotary solutions. Data center operators must look to design for thermal management, cooling performance and energy transmission efficiency in the next 18 months to manage and limit the predictable high density power and cooling costs.”

Design issues. The aging data center stock brought fiber and total network/transport layer design issues to the fore. Worn says, “Fiber management represents a minor portion of the capital outlay in a datacenter, in fact the total cost can be less that 3%, however the role of this fundamental medium is critical to the reliability and availability of operational high capacity systems, and with ever increasing server densities fiber design, implementation and management is even more mission critical. From poor installation, the limitation in this medium, to rushed ongoing commissioning proper fiber management, the event highlighted the role of ensuring fiber integrity in terms of getting business to zero downtime to protect the life blood of the enterprise – data.”