– Mark Harris, vice president of marketing with Server Technology (www.servertech.com), says:
Data Center power has become a highly visible topic of discussion across the industry. Power is being considered by many as a truly Mission-Critical resource rather than a common facility service. Today’s data centers are becoming higher density and consuming levels of power like never seen before. Whereas a rack used to consume a few kilowatts, we are now seeing five to ten TIMES that amount of power per rack routinely. What that means is that the choices for power delivery must be made much more carefully than ever before, looking for vendors that have ‘done their homework’. At Server Technology we have authored nearly two dozen patents associated with building rack based PDU products and have huge investments in climate-controlled stress chambers and power stress apparatus to assure that our products continue to delivery power over their entire useful lifetime without failure. Server Technology has been in the data center power delivery business for over 25 years and has amassed a wealth of experience creating mission-critical rack based power delivery appliances. This is a key point that data center managers should look for, products that are DESIGNED to run under stress conditions from vendors that are LEADERS and have the experience and reputation of doing what they say under extreme circumstances. Frankly it is quite easy for any vendor that wishes to build a PDU product that works on the FIRST day, but it is MUCH harder to build the same kind of product that does so for a couple THOUSAND days or more.
Until recently, data center power considerations were relegated to the facilities organizations and were not considered critical path. It has become apparent over the past few years that all of the amazing (and power hungry) innovations being made by the server, network and storage vendors run the risk of failure when powered by anything other than mission-critical infrastructures. Like Lasik for your eyes, most data center managers are realizing that no one benefits over the long term by choosing the ‘cheapest’ PDU on the market. Failures in modern data centers change peoples’ lives forever. Everything along the power chain needs to be assessed or re-assessed for suitability in a modern data center filled with much more hungry racks housing much higher density computing. Today, many Data Center operators consider the delivery of power (and the associated cooling) to be their HIGHEST priority concern when building or retro-fitting new data center facilities, while others are still trying to convince their business managers that investments in the physical infrastructure are more important that looking for the next version of a server or switch.
The modern data center manager or operator and those professionals looking to outfit data centers, should consider building facilities that can operate over long periods of time, allowing for a multitude of dense racks consuming 20 kilowatts or more (high power delivery), and assure that the required level of cooling would be available and consistent with their deployment needs. That said, the traditional approach of over-provisioning in the data center is a business practice that has fallen by the wayside and has been replaced by a dynamic or modular approach. The ultimate goal for a data center manager is to closely align processing capacity to user demand. No more, no less. Data centers should have the CPU cycles available on demand, and the cooling that goes with it. To do so, power distribution must be outfitted in such a way to allow smooth operations in times of light or heavy demand. The power chain must be actively managed, metrics being looked at for quality and performance, and potential problems identified BEFORE they result in failures. Consider investing in the technologies which allow for this wide range of performance needs (dynamic provisioning of power, cooling and processing) over the useful lifetime of the data center.