In his “Zinc Whiskers” column in the January/February edition of Mission Critical, Bruce Myatt discussed the recent history of containerized-modular data centers (C-MDC), the major players in the industry, the benefits of standardization, and the industry skeptics who remain unconvinced. Elsewhere in this issue (see p. 32), Ramez Naguib of Harsco Industrial talks about modular cooling plants, the ability to more confidently match capacity to loads, and some of the financial benefits of going C-MDC. Both industry experts discuss the growing number of applications for C-MDC solutions.
Some of the factors that explain the growing popularity of C-MDC include quality control, increased efficiency, superior monitoring, cost, and schedule. It’s apparent that C-MDC solutions will be part of the landscape, at least for the immediate future, and it’s equally obvious that data center owners/operators must know what to look for when considering a C-MDC solution.
Ever since the first data centers were built, the challenge of coordinating multiple trades, facility groups, and IT groups has hampered quality control and speed to delivery. Making matters worse, the trade and facility groups often lacked understanding of the technology, and IT did not understand power and cooling.
Much has changed in the last two decades. Increasingly IT and facilities have developed a better understanding of today’s technology needs-in many cases evolving into a single group. C-MDC is the next step in this evolution as IT containers or modules can be delivered complete (single sourced) with power, cooling, communications cabling, hardware, and software all built in a factory-controlled environment. Gone are the days of split responsibilities, confused integration, and uncertain performance expectations. Factory-built systems have the advantages of repeatability, proven performance metrics, standardization of design sections, and significantly less field labor.
With both the cost and demand for power increasing, energy efficiency is all the rage in data centers today. Manufacturers are providing repeatable tested performance metrics on their C-MDC products so that IT and management can reliably calculate their PUEs and carbon footprints and calculate how it will change. By contrast, in many traditional data centers, operators are still figuring out where to measure performance and what system to install to record it.
Over the years, the industry has spent millions on monitoring. Often data center monitoring is not addressed until the design is done and even then is usually “valued engineered” later to bring the project in on budget. Further, there is always great debate as to what to monitor, how to monitor it, and even whether anyone will ever look at the data. Historically these monitoring systems have run under vendor proprietary software and only recently have become more cross-vendor friendly. Most C-MDC vendors offer a C-MDC package complete with hardware/software and graphic displays that present real-time values for every operating component. Their systems are fully compatible with all the major IT and facilities systems, are IP addressable, and provide live operating metrics like PUE and carbon footprints.
There is much debate about the cost effectiveness of C-MDC. How expensive are they compared to a traditional data center? How does one compare the cost of a C-MDC solution to a more traditional solution?
Some argue that C-MDC more closely matches costs with immediate needs; that is, all the expansion space does not have to be built day one. C-MDC units can be added as needed in the future. So unrealized projected growth doesn’t cost anything. The traditionalists, however, counter that in the end C-MDC costs more, compared to a traditional data center that is rapidly populated.
Technology continues to change. A C-MDC unit rated for 5 kilowatts/cabinet (kW) might suffice today until an enterprise needs to upgrade to 15-kW/cabinet units in five years, deferring the purchase of high-powered infrastructure until that time and avoiding five years of cost associated with excess capacity. Predictions about capacity needs in data centers never seem to be adequate. What can be stated about C-MDC costs is that they have come down dramatically as production and competition ramps up.
Schedule is where C-MDC usually shines, particularly when considering total schedule and not just the construction schedule. A traditional data center cycle sequence includes a planning-and-design cycle, construction, cabling, hardware deployment, software, and data burn-in. The planning-and-design cycle of a C-MDC is shortened because standard products have known repeatable requirements.