How to Choose the Right Cloud
A closer look at the three types of cloud computing - private, public, and hybrid
To date, much of the discussion surrounding the cloud has centered around "how do we build it?" or "how do we access it?" Very few people have asked "what do we do with it?"
The simple answer, of course, is that it will expand our resource and service options, but that only scratches the surface of what the cloud ultimately represents to data center infrastructure. Like a lump of clay, the cloud can be molded into a work of art, or a pile of rubble. The decisions made at the outset of any cloud program could vastly affect its efficacy in real-world situations, so it helps to know what we hope to gain from this technology at the outset.
Different types of clouds support different business models, according to communications consultant Gary Audin. Some, like Amazon's EC2, are pure cloud plays, in which cloud-based resources deliver traditionally licensed software or their own or third-party services. Others have gone for complete vertical integration in which the cloud provider owns the entire infrastructure - hardware, software, networking and the staff to run them - and assumes full responsibility for service levels, security, availability and the like. These and other models can differ greatly in the type and scale of provided services, providing varying levels of support for specific data functions and business processes.
This still leaves us with the question of how we should implement cloud services, however. While specifics will be as varied as the number of enterprise applications available today, some broad outlines are already emerging. Researchers at IBM's Institute for Business Value have identified three main approaches to cloud computing as means to impact organizational and customer-facing value propositions. Optimizers will favor incremental enhancements to improve efficiency, while innovators will aim for new revenue streams and more substantial role changes through the cloud. Finally, disruptors will look for entirely new value propositions through the creation of cloud-based customer needs, segments and value changes. This isn't a progression, mind you, as many organizations will fit comfortably in one or more roles for extended periods.
It is also very likely that this variety will extend to underlying cloud infrastructure before too long. As InformationWeek's Charles Babcock pointed out recently, telecommunications providers like CenturyLink and Verizon are itching to move in on top cloud providers like Google and Amazon, and are sitting on networks that could bypass the Internet completely in favor of highly secure, WAN-based services. Through acquisitions like Verizon's recent buy-out of Terremark and CenturyLink's of Savvis, there is a distinct possibility that we could soon see highly integrated chains of cloud-based data centers on networks that are entirely separate from public infrastructure.
And despite what you hear about all the wonderful things it can do, the fact is that without proper infrastructure and application management, the cloud can just as easily turn into another unproductive cost center, says Bluehawk Networks' Dave Saunders. Now that responsibility for data services and availability are shifting away from traditional IT toward the user population, enterprises should begin to ask key questions regarding the creation, collection and use of data. The answers may lead to some radical new thinking over infrastructure development and support, asset management and workflow direction.
The cloud is no exception. Without a solid set of expectations and a realistic view of what it can and cannot do, the cloud will simply serve as the extended version of today's data infrastructure.