I had the privilege of keynoting a data center conference in Moscow this week, speaking on the topic of what cloud computing means to the data center of the future. This is the largest data center conference in Russia, and attracts a mix of internal data center facilities executives as well as hosting providers. The conference was extremely well-attended, with record numbers of people registering.
[For timely cloud computing news and expert analysis, see CIO.com's Cloud Computing Drilldown section. ]
As background, Russia is coming out of a severe financial dislocation; its economy shrank 7.9 percent in 2009. It appears to be on a growth track today. Furthermore, its penetration of IT use in the economy is far lower than what we see in the U.S. and western Europe - it can be characterized as an emerging economy, though, if my observations are anything to go by, one with tremendous vitality and ambition.
What that means to the future of data centers is quite intriguing. Unlike the U.S., for example, a much lower percentage of companies in Russia have significant existing data center infrastructure. For many companies, the choice of computing infrastructure presents an interesting greenfield opportunity. The question, therefore, is what path to pursue as companies build out their infrastructure of the future.
A complicating factor is that Russia suffers the challenges of developing economies: reliable electricity and Internet connectivity - although I must say that at the Radisson I am staying at, the (free!) Wi-Fi delivers 13 Mbs down and 9 (!) Mbs up, well beyond what I've ever gotten at any U.S. hotel, at any price. Nevertheless, the power and connectivity challenges make running data centers difficult.
Those challenges argue for keeping a company's data center as close as practicable, to reduce the potential for service interruption. On the other hand, the cost of building a data center has skyrocketed over the past few years, because the sophistication and density of computing infrastructures has increased significantly.
Slideshow: Secrets of Successful Data Centers
These factors make it difficult to know which alternative makes the most sense:
Build an on-premise data center. This provides less potential for service interruption due to connectivity problems, and, should power go out in some other location in the area, enable the entity to continue compute activities. It does, however, commit a company residing in a less-developed economy to an expensive capital investment and an ongoing capital expenditure regime, as equipment wears out or becomes obsolete and requires replacement.
Leverage a hosting provider. This avoids making a commitment to building and maintaining a compute infrastructure, while still allowing a company to leverage a state-of-the-art data center environment. Moreover, it allows a company to avoid hiring at least a portion of the operations staff required by the previous option. It also offloads a capital expense onto another party, freeing up capital for investment in other, perhaps more business-focused, areas. This last item should not be minimized, given the common perspective I heard voiced that Russia provides tremendous business opportunities right now, and those that can jump into the lead may cement a permanent advantage. The ability to direct capital toward production capacity, store building, working capital, or even hiring additional customer-facing personnel is no small thing in such an environment.
This is a challenging dilemma, and like all dilemmas, presents itself with no clear cut answer. Making decisions with ambiguous information is one of the most difficult and emotionally daunting situations one can face.
As an input to the decision, the hosting business in Russia is going gangbusters. Many companies, perhaps needing compute capacity sooner than could be made available by on-premise construction, are flocking to providers who have infrastructure in place.
For what it's worth, I believe that Russia will ultimately tilt toward external cloud providers, whether so-called public providers like Amazon, or private cloud providers akin to Terremark or BlueLock. The absence of a sunk cost existing infrastructure means the decision about internal vs. external is less skewed by existing practices or an employee base, threatened by potential job loss, lobbying for keeping an on-premise solution. If one examines a choice between making a significant investment that imposes a delay prior to availability because of construction timeframes and making an opex payment for immediate capacity, the decision seems much more obvious than that same choice made in an environment like the U.S., in which on-premise compute infrastructure is probably already in place.
I predict that Russia, like other developing economies, may pursue cloud computing in much the same fashion as they have pursued telephony. Because most developing nations have no robust landline infrastructure, when they come to putting telephony into place, they completely skip the capital-intensive, construction-heavy landline stage of telephony altogether. They go directly from low telephone penetration to massive mobile penetration without any attempt to put fixed telephony into service.
This is not to say that nations like the U.S. should not have put copper in the ground. Far from it. The advantages of easy communication are so great that they justified the capital investment of landline telephony. There was no alternative to copper investment at the time the U.S. and its sister developed nations made those investments (of course, at that time the U.S. was nearer developing nation status; despite its lower economic state, it still made sense to sacrifice for communication, so great were its benefits). It's just that nations implementing telephony today have other, more flexible, less expensive options that make greater sense than landline investment.
The move to cloud computing will offer the same efficient economic choice. If one didn't need to make a huge investment in infrastructure to gain the benefits of computing, would one make that decision? The answer is quite clear. The entire goal of constructing a data center is to gain access to compute capacity (many would make an even more stringent purpose analysis, saying the entire goal is to gain access to application functionality, which powers business objectives). If one can gain access to compute capacity absent a large capital investment, why would precious capital be devoted to computer hardware?
Far more likely for emerging nations is that individual companies will choose to use external compute capacity, while the countries themselves will focus on putting fiber-optic connectivity infrastructure into place to support the new model of computing. In a sense, emerging economies are blessed by coming to the computing game late; they can move directly to more effective approaches and perhaps be even more nimble than early adopters.
Of course, as Adrian Monk often observes, something can be both a blessing and a curse. The move to use external cloud computing imposes the need for crisp application architectures and security practices, elements that could be elided when computing was done on-premise and assumptions made about the protection offered by a data center perimeter. Being forced to adopt better practices in architecture and security is actually a second blessing, difficult though the adoption may be in practice.
The future of computing is partitioned, cleanly-interfaced, highly scalable applications with security focused on an endpoint approach. Using an external cloud merely guides organizations toward those practices more quickly, paying further benefits in application extensibility and integration capability.
It's been quite a visit to Russia and I thoroughly enjoyed my exposure to the exciting, sprawling, dynamic society. I believe the future of Russia's compute environment will be just as lively.
Bernard Golden is CEO of consulting firm HyperStratus, which specializes in virtualization, cloud computing and related issues. He is also the author of "Virtualization for Dummies," the best-selling book on virtualization to date.
Read more about virtualization in CIO's Virtualization Drilldown.