You Say You Want a Cloud Revolution

structure_speaker_seriesTake yourself back for a moment to 1990, to the era of dueling operating systems: OS/2 and Windows. At the time, many people still used MS-DOS, and Windows was new (and klunky). Microsoft had cooperated with IBM to create OS/2 to overcome the limitations of DOS by adding multitasking, protected mode, and enhanced video APIs. OS/2, they both trumpeted, was a revolutionary computing platform.

Oops. Guess what? Turns out no one wanted revolutionary. We all wanted those improvements, to be sure, but we wanted them delivered in a way that didn’t require redesigning and rewriting our applications, or limiting the devices we could use. Voila! Windows 3.0 brought us evolutionary OS advances, and we all know who won.

What does this have to do with cloud computing? Well, the same principle applies to cloud offerings today. The easier a platform or service is to adopt for existing applications and uses, the more popular it’s going to be, whereas the more it breaks with current practice, the less widespread its appeal.

Take Amazon’sEC2. A key part of its success has stemmed from a design that represents an evolutionary (not revolutionary) advance in IT deployment architecture. EC2 instances — virtual servers — run identically in most respects to servers in your data center. The major differences lie in how you acquire and pay for them. In that sense, EC2 represents a revolutionary business innovation coupled with an evolutionary technical innovation. The technical differences in EC2 instance behavior –- for instance, transient local disk storage — are small enough that developers can migrate applications with minimal changes. In order to achieve fully functioning deployments, additional design and operational capabilities have to be added (familiar territory to us at RightScale). But the starting point is pretty familiar.

Contrast that with another Amazon cloud service that’s more revolutionary in nature. It’s called SimpleDB, and it provides an expandable, fast data store that addresses the problem of scalability common to relational databases. The catch is that it does so at the cost of incompatibility with some common relational database functions, while adding traits that represent a big enough departure in design to limit the service’s appeal. In other words, it’s too revolutionary. From my observations, use of traditional relational databases in the cloud completely trumps usage of simple data stores.

Michael-CrandellIt’s interesting to review the list of cloud services — including those offered by Google, Sun, Salesforce, Rackspace, GoGrid, Eucalpytus, VMware and Amazon — and rate them on the evolutionary/revolutionary scale. As an exercise, it reveals how easy it is for customers to start using these services effectively and, therefore, how quickly they will expand the cloud market.

That is not to say that revolutionary technology solutions aren’t useful or valuable. They are. But they’re unlikely to drive substantial market growth in the near term. Not surprisingly, some of the bigger hurdles to cloud adoption at present revolve not around cloud infrastructure technology, strictly speaking, but around departures from traditional data center models that are still perceived as too revolutionary from a business perspective. Security, governance and compliance are probably the most prominent examples, as detailed here, although that is changing fast.

At this stage in the development of the cloud computing market, the biggest opportunity lies in helping companies make evolutionary shifts to cloud architectures. That’s the low-hanging fruit for both customers and vendors alike. Relatively quick, easy success stories are the ones that stand out in the cloud market today, and are driving the business. The reason is simple: If you can take existing IT requirements and fulfill them with a system that is both more agile and less expensive but doesn’t require major changes to your applications, that’s revolutionary.

Michael Crandell is the CEO of RightScale Inc.

loading

Comments have been disabled for this post