With companies relying on IT more heavily than ever, data center capacity requirements are steadily rising. Unfortunately, so are the costs associated with data center construction and operation. As a result, organizations are increasingly searching for ways to reduce the size of their data centers without compromising their ability to meet business requirements.
The good news is that companies interested in following the trend toward smaller data centers can employ a wide range of techniques to shrink their computing footprint. IT consolidation is best approached by a combination of three primary strategies, design-level, white space and grey space. The following will explore four design-level strategies that can help you make the most out of your available footprint.
1. Tailor your power architecture to mission requirements: A 2N power architecture features dual, identical power paths all the way from utility main to load. Typically, each path operates at no more than 50 percent of capacity, so if either one becomes unavailable due to a utility service interruption or required maintenance, the other one can keep the data center fully operational. Generally speaking, 2N power architectures are a wise choice for organizations with an extremely low tolerance for downtime, such as banks, utilities and emergency service providers.
As you would expect, deploying two of everything takes up a great deal of floor space. Companies with more mainstream uptime requirements can reduce their data center footprint while still keeping availability levels high by using an N+1 power architecture instead of a 2N configuration. An N+1 architecture includes one more of each power quality device than the minimum required to keep the facility up and running. That way, if any one device experiences an outage or needs maintenance, the remaining systems can still protect servers from data loss. This is a good application for redundant UPSs as well as redundant rack power strips and remote power panels (RPPs). N+1 configurations require far less space than 2N architectures, yet still provide adequate redundancy for most purposes, according to The Uptime Institute.
Looking to compact their footprint even further, large data center operators such as Amazon and Google are increasingly using N and even N-1 power architectures. An N architecture doesn’t include redundant uninterruptible power system (UPS) hardware, leaving critical workloads potentially unprotected in the event of UPS failure and during maintenance procedures. However, organizations that employ N architectures typically utilize state-of-the-art data mirroring techniques to synchronize multiple data centers on separate power grids. Should downtime occur at one facility, another one can keep mission-critical systems available to users. N-1 architectures take the same logic a step further by employing no UPS hardware whatsoever. Companies that use such configurations count on the availability of redundant facilities to compensate for the significantly higher likelihood of any one facility suffering data loss.
Though they dramatically lower data center floor space requirements, both N and N-1 power architectures are generally safe options only for the largest and most sophisticated data center operators. Furthermore, both architectures make sense only for companies with enough funds to build and maintain multiple data centers, even if one facility is all they really need.
2. Utilize 400V Power: In the U.S., utilities typically deliver power at 480V. Most U.S. data centers, however, operate at 120V/208V. As a result, they must use a series of transformers to “step down” power from the 480V at which it’s received to the 120V at which it’s consumed by servers and other infrastructure devices.
Operating a data center at 400V, instead of 120V/208V, reduces the number of transformations that must occur along the power chain. That, in turn, can lower floor space requirements by up to 60 percent, by eliminating the conventional PDU transformers. Further space can then be saved by utilizing busways for power distribution or 400/230V power. Manufacturers now offer UL-listed 400V UPS, RPP, busway and rack power devices.
As an added benefit, 400V data centers tend to be more energy-efficient as well, since fewer transformations generally translates into less power waste, and servers operate more efficiently at 240V as opposed to 120V.
3. Deploy containerized data center modules: Seeking to pack more power and computing capacity into less space, data center operators are increasingly utilizing containerized infrastructure modules. Some such products, from equipment manufacturers such as Oracle, HP, Dell and IBM, feature ready-to-use server racks. Others, such as the Eaton Uninterruptible Power Center, provide modular, pre-configured UPS systems. Yet another set of offerings include both the server hardware and associated power and cooling needed to run a miniature, self-contained computing facility. In addition to dramatically reducing the time required to expand or reconfigure an existing data center, all three module types enable companies to grow a data center up instead of out, shrinking its physical footprint by stacking modules on top of one another.
4. Employ a “dark” data center design: To reduce IT operating costs, some companies are experimenting with dark, or unmanned, data centers. Such facilities rely on automation and remote management software in place of onsite technicians. Well suited to disaster recovery sites that get little regular use, dark data centers generally cost less to run than traditional ones. In addition, they also tend to be physically smaller, as they require significantly less office space and fewer parking spots than manned facilities.
Closing Thoughts: Today’s IT and facilities managers face a difficult bind: Though computing needs are constantly escalating, so is the cost per square foot of data center space and the price of critical supporting resources such as electricity and water.
Consequently, more and more businesses are attempting to shrink the size of their data center without also shrinking its operating capacity.
Fortunately, there are many ways to compact a data center while still meeting business requirements. Most of them, moreover, are proven and cost-effective options for companies of almost any size. By following the design-level recommendations outlined above, you can get more done in less space and position your company to meet its IT requirements with maximum cost effectiveness.