Veteran industry analyst Judith Hurwitz astutely compares data centers to garages. When you buy a house, there's plenty of room for two cars. But over the years, the clutter builds up, and parking even one car becomes a challenge.
So too with data centers. The original motivation for these facilities was to host systems of record: those workhorses that run the core processes of the enterprise. But over time, we cluttered our data centers with an increasing variety of systems, which increased costs and risked impacting the systems of record that justified the data centers in the first place.
Enter Cloud Computing. Now we can move all that garage detritus, also known as systems of engagement, to the Cloud. Customer-facing apps, CRM apps, and the like run just fine in the Cloud, leaving the data centers to their original purpose: hosting the systems of record.
This prediction contradicts our prediction that in the future, all data centers will be Clouds. But perhaps Hurwitz has identified an exception. Why include virtualization, dynamic provisioning, and automated configuration and management in a data center when all it's doing is running stable, dull systems of record? Uptime is more important than elasticity for these systems, after all, and uptime is what properly configured data centers have always been good at. Maybe such data centers will never be Clouds?
There are two weaknesses to this argument. First, the line between dynamic systems of engagement that can leverage the agility benefits of the Cloud, and stable systems of record that simply need to keep plugging away continues to shift, shrinking the role of the systems of record. Sure, there will always be some core transaction processing going on under the covers in industries such as banking and insurance, but even these stodgy industries are becoming increasingly dynamic and customer-focused. So maybe we do need a clean garage, but for an increasingly smaller car.
Second, while elasticity is perhaps the most important essential characteristic of the Cloud, it's not the only one. Extreme durability is another. Today's Clouds are perhaps too immature to deliver high levels of durability, but products like Amazon's Glacier indicate the direction this part of the Cloud marketplace is heading. So, once we work out the kinks in high durability Clouds, then why wouldn't any CIO want to host their systems of record in such environments? We're not there yet, but it's just a matter of time.