10 Steps to Increasing Data Center Efficiency and Availability
A systematic approach to building data center infrastructure management.
The modular data center movement recently passed the 5-year mark, and it's fair to say that results so far have been mixed.
After all, no one really expected much when Sun released the first fruits of Project Blackbox, a 20-foot containerized data environment that the company said could be up and running for about one-hundredth the cost of a traditional data center. Conventional thinking, then and now, holds that modularity is suitable for niche deployments or quick expansion of data resources, but not as a standalone data infrastructure.
Sun, of course, has since been acquired by Oracle, which by all indications seems intent on continuing the development of modular systems. So it must see at least the potential for future growth, particularly as the expected legions of new cloud providers look to ramp up infrastructure in the drive to deliver top-notch services with the lowest possible overhead.
According to a recent survey by Data Center Knowledge, nearly two-thirds of traditional brick-and-mortar enterprises have no plans to deploy modular solutions at the moment. However, the third that remains still represents a sizable market that is more than capable of producing the level of revenue needed to support further research and development. And it's worth noting that Sun has been joined by HP, Dell, IO Data Centers and others who cater to such deep-pocketed clients as Microsoft and Amazon.
There's also the fact that data services are starting to permeate previously unwired regions of the globe. Africa, Central and Eastern Europe, and large portions of Asia and South America are clamoring for digital infrastructure but are unwilling and/or unable to pay top dollar for traditional data center infrastructure. Modular solutions are tailor-made for users who need quick and easy deployment in areas where support infrastructure like electricity and running water may be lacking.
In India, for example, IBM recently popped up a 900,000-square-foot facility for Tulip Telecom Ltd. in only 10 months. The facility consists of four structures packed with 20 Enterprise Modular Data Centers and is designed for IaaS operations using the IBM SmartCloud platform. It has been designed with redundant power and networking to boost availability and has been given a very respectable 1.5 power usage effectiveness (PUE) rating.
To all the established enterprises out there, I pose a simple question: If you had to do it all over again, would you build your legacy architecture from scratch? Be honest.