Time for a Serious Look at Modular IT

Slide Show

A Closer Look at Converged Infrastructure Adoption

Today’s enterprise is caught in a conundrum born of the need to cut costs in a time of increasing data loads. Big Data, rich media content, and the increase in mobile traffic are all putting pressure on data resources that for the most part were designed for the static architectures of a bygone era.

Virtualization and cloud computing have helped carry this burden to a point, but a single piece of hardware can host only so many virtual machines, and the cloud provides the kind of scalability needed for today’s loads only when public resources are brought to bear.

Small wonder, then, that modular solutions, or converged infrastructure, are starting to make their presence known in enterprise circles. But while there’s no doubt that modular systems are cheaper to buy and implement, the question is: How much cheaper? And in the end, is this technology any more scalable, flexible and most importantly, reliable, than traditional IT infrastructure?

On the cost side at least, some hard data is finally coming through. According to DCD Intelligence, modular systems average about 14 percent less to deploy than traditional build-outs. The caveat, though, is that with such a wide variety of solutions on the market and an even wider variety of bricks-and-mortar data infrastructure to compare them to, actual results will vary widely from deployment to deployment. As the study’s lead author Chris Drake notes, higher density and more efficient use of space in modular architectures will almost certainly improve efficiency, but the sheer number of standards, certifications and levels of customer support makes a true apples-to-apples comparison virtually impossible.

One thing that cannot be denied, however, is the growing popularity of modular technology. MarketsandMarkets reports that the industry is on pace to hit $40.41 billion in revenue by 2018, which gives it an average yearly growth rate of about 37.14 percent between now and then. And the company notes that this is still the introductory period for modular systems. Once the enterprise industry becomes more familiar with the concepts and functionality of the technology, expect growth to accelerate even faster into the next decade.

Modular systems also dovetail nicely with the other major trend sweeping the data center: green IT.  Hyperscale users like eBay in particular are under intense pressure to cut their power consumption, and the more efficient use of power and cooling resources that modular technology provides is a welcome benefit. The eBay Project Mercury plant in Phoenix, Ariz., for example, has achieved a power usage effectiveness (PUE) of 1.2, which is only a hair above a perfect score of 1.0. The facility, designed by modular systems contractor IO, utilizes the company’s pod approach, which combines data hardware and software, as well as power distribution equipment. It is intended primarily for core back-end infrastructure, and individual pods can be up and running in as little as four months.

Modular solutions have another key advantage over traditional data infrastructure, particularly for coastal facilities looking at another hurricane season: They can be built to better withstand extreme weather conditions, or even be placed on a mobile footing and relocated if the situation warrants. Over the summer, the Georgia Port Authority unveiled its Mobile Command Center, a 53-foot, truck-mounted facility designed to step in should primary resources be lost. The center holds two separate data facilities: one for data operations and the other for security functions, such as video surveillance and emergency alert capabilities.

Sure, modular technology has a lot going for it as the enterprise strives to become leaner and meaner. Traditional data center infrastructure represents a substantial investment that will undoubtedly be leveraged to the greatest extent possible, but when it comes to scaling up and out to accommodate new private clouds and the demands of increasing data loads, modular systems will be hard to ignore, either as primary systems or adjuncts to legacy infrastructure.