Virtualization, cloud computing, big data, and converged infrastructures are rapidly transforming corporate information technology (IT). Unfortunately, the data centers many businesses rely on at present lack the power and cooling capacity to handle those technologies. Worse yet, the packaged rack-based power and cooling solutions that some vendors claim increase the efficiency and power density of data centers can be both costly and inflexible.
By designing a solution that utilizes best-of-breed power and cooling systems, today’s data centers can maximize capacity and minimize waste without becoming locked to a limited set of deployment options and vendors. Sophisticated uninterruptible power system (UPS) hardware, intelligent power distribution, flexible and high-efficiency containment, and robust power management solutions are available today that can enhance efficiency and power density more completely and cost-effectively than traditional, packaged power and cooling solutions — helping data centers of any size manage growing demand in a strategic fashion.
Today’s high-density IT solutions, including cloud computing, virtualization, and converged infrastructure solutions, decrease IT overhead and increase business agility. However, most data centers in operation today — and especially those more than eight years old — have trouble accommodating these powerful new technologies. In addition to operating inefficiently, obstacles include inadequate power density, power and cooling capacity, power and cooling redundancy, management capabilities, and rack deployment flexibility.
THE OLD SOLUTION
To help businesses implement power- and cooling-hungry technologies more quickly, several vendors have introduced packaged power and cooling upgrade solutions that include most of the components needed to increase an existing data center’s efficiency and density. However, these solutions have substantial limitations.
Packaged solutions can limit cooling, positioning, and thermal management options and often lack robust management and monitoring functionality. The standardized “one pod fits all” approach can also lead to stranded capacity. Network, storage, and server environments have three distinct load density requirements, so a singular package design does not work for all. For example, packaged solutions usually favor one particular cooling deployment mode, regardless of an individual data center’s structural constraints and other limitations imposed by the building’s original design.
In addition, packaged solutions typically overuse in-row cooling. While effective when used appropriately, this system can impose steep purchase and deployment costs, and take up valuable floor space that could otherwise be dedicated to IT equipment.
Purchasing power and cooling systems from a single vendor can save money and simplify maintenance but may limit your freedom to add UPS, PDU, and containment products from other vendors in the future as your needs and preferences change.
THE NEW SOLUTION
Instead of looking toward a particular packaged solution, data center managers should ensure each component specified for the data center is energy efficient, flexible, scalable, and highly manageable prior to deployment.
For example, modern containment solutions and late-model UPSs dramatically improve energy efficiency, enabling companies to save significantly on power. Advanced UPSs currently available can routinely deliver over 97% efficiency in double conversion mode and achieve more than 99% efficiency without sacrifices in reliability. And, when coupled with intelligent management software, modern power distribution units provide branch circuit monitoring functionality that allows you to track power usage all the way to individual servers. That’s detailed information you can use to spot trends and eliminate waste.
Further, high-quality and strategic containment solutions can be designed to meet a specific facility’s needs and raise efficiency by minimizing the intermingling of hot exhaust air with cool supply air. In fact, top-of-the-line offerings cap supply air leakage at no more than 3%, and limit maximum gap leakage to no more than 3% of total surface area. Moreover, truly trustworthy containment vendors will verify that they’ve delivered on their air leakage commitments by taking detailed post-deployment measurements at your data center.
In addition to increases in efficiency, handpicking best-of-breed products can lead toward the increased flexibility needed to cost-effectively add capacity down the road. For example, facilities that use modular UPS systems and busways rather than cables can gradually and cost-effectively add capacity as changing needs dictate. Today’s busway systems are easier to configure than cables and easier to modify or expand as changing needs dictate, too. Modern busway solutions feature distribution boxes known as bus plugs that include a circuit breaker, one or more receptacles, terminal blocks, or even pigtail cables with receptacles that can be dropped directly into the racks below. As power density needs change, technicians can re-position bus plugs at convenient positions along the length of the bus as necessary to ensure that each one continues to feed one rack or several adjacent ones. That enables data centers to perform moves, additions, and changes to their rack layout faster and more flexibly without dedicating valuable floor space to remote power panels or enduring the safety concerns those units raise.
Organizations can also place modern UPS hardware at the end of a row, in the center of a row, against the wall, in a corner of the server room or back-to-back, with front access required only for service and installation. Additionally, today’s best-designed power systems can accommodate either raised floor or overhead-wired installations, and come with auxiliary cabinetry that can be placed on either side of the UPS cabinet. Further, these modern UPS solutions can be incorporated within the data center’s comprehensive heat containment system. In turn, improving reliability by permitting UPSs to reside in close proximity to the loads they protect.
To meet future needs easily, UPSs now allow you to add capacity in 50 kW increments to a maximum of 200 kW. That frees companies from buying more capacity than they need upfront and provides increased redundancy at a far lower cost in money and floor space than buying and deploying a second UPS and battery.
Finally, by deploying best-of-breed power management solutions, today’s data centers can achieve the power monitoring capabilities needed to protect and administer virtualized environments. Software is as important to raising data center efficiency as hardware, so make certain to deploy software that can do the following:
Provide unified administrative control. Advanced power management systems also enhance the capabilities of virtualization management products such as VMware’s vCenter™ Server by enabling data center personnel to view, monitor, and control not only physical and virtual servers but UPSs, PDUs, and other power devices through one interface.
Use an agentless way to connect power management to the virtualization management console. Traditional power management applies a single agent per virtual machine. This does not scale when many virtual machines can be spun up and down quickly and agents need to be individually deployed and updated. There is also an input/output bottleneck as all agents simultaneously respond to a power event. Using the direct hook to the virtualization management console from the power management software removes these issues at a stroke.
Perform live virtual machine migrations. Drawing on seamless integration with live migration systems such as VMware’s vMotion™, best-in-class power management solutions can automatically and transparently move virtual machines impacted by a power outage to unaffected servers elsewhere on the network or in a co-located cloud data center. Additionally, they can use that same live migration capability to optimize power efficiency by automatically re-distributing virtual workloads from enclosures nearing their power or cooling limit to racks with spare capacity.
Aggregate power protection and distribution device information. Leading-edge management solutions give IT and facilities administrators a truly global view of their power protection and distribution infrastructure through a single console by automatically discovering and collecting real-time data from network-enabled power devices both inside and outside the data center.
Protect workloads flexibly, and initiate system shutdown during power outages. For those data center managers without standby generators, intelligent power management solutions help technicians extend UPS battery life by grouping their least important infrastructure resources into distinct load segments that can be shut down first after a power outage. During extended utility interruptions, next-generation power management solutions can also protect unsaved work and preserve data integrity by shutting down affected servers and network devices both automatically and gracefully.
Boosting an existing data center’s efficiency and power density with best-of-breed components is faster, easier, and more economical than building a new facility and much less constraining than using a packaged power and cooling solution. In particular, companies that utilize the best-of-breed power and cooling components recommended above can expect to collect significant returns on their investment.
By increasing overall power and cooling efficiency, the efficiency and power density upgrade solutions outlined can enable companies to achieve more capacity from their existing power and cooling systems, delaying the need for costly upgrades and replacements. Implementing efficiency and density upgrades also extends the life of a data center, enabling organizations to get more use from existing facilities and defer the substantial expense of constructing new facilities.
Virtualization, cloud computing, big data, and converged infrastructures help companies substantially lower their costs and dramatically improve their competitiveness. But, before organizations can realize those gains, they must prepare their data centers to satisfy far more intense efficiency and power density demands. Data centers equipped for maximum efficiency and density will have the scalability needed to reliably address today’s heightened requirements, and future needs.