Manage Learn to apply best practices and optimize your operations.

Experiments with hybrid clouds give life to Frankenclouds

As more enterprises move to hybrid clouds -- usually combining AWS public cloud and on-premises VMware -- some IT shops are creating environments with two, three and even four cloud platforms. And they're even learning to love their Frankenclouds.

Stories about enterprises that do an elegant wholesale migration to Amazon Web Services come along every once in a while. Conventional wisdom, however, states that hybrid clouds -- usually a combination of AWS public cloud and on-premises VMware -- are where the world is headed. More recently, I've been talking to IT shops with not just two, but three and sometimes four distinct "cloud" environments. These are no longer hybrid clouds -- they're Frankenclouds.

When talking to IT teams about enterprise DevOps shops, one of the many of the interesting themes that emerged was that, more often than not, IT pros run workloads not only on AWS and VMware, but also alongside OpenStack or CloudStack, some bare-metal legacy kit and, most likely, some form of platform as a service.

The Canadian Yellow Pages Group, for instance, runs its digital directory services across AWS, VMware, OpenStack and Google App Engine, and also relies on RightScale to provide commonality between the platforms. Deutsche Telekom HBS delivers its small-business IP telephony and collaboration services from a VMware and CloudStack private cloud, but uses AWS for its test and development environments with the help of Ravello.

The reasons for creating these complex, multi-headed cloud monsters are as unique as the companies that put them together: legacy infrastructure and technical debt, cost, feature and functionality needs, data sovereignty requirements, internal skillsets. In short, all the things you would expect.

Much like Dr. Frankenstein learned to accept his successes and failures with quiet dignity and grace, IT professionals work in these complex environments with a surprising degree of equanimity. That's because they use configuration management systems to produce automated infrastructure as code, be it on AWS, GCE, or RackSpace, VMware, OpenStack or Eucalyptus. They then test that infrastructure as part of a continuous integration and delivery process. The result is that the underlying infrastructure is sort of beside the point.

"Infrastructure as code makes it so that you're not sentimental about it anymore," said Jonathan Chauncey, an R&D engineer at Rally Software, whose agile-inspired project management service runs on a combination of AWS and VMware. "It's just about getting bits on a box."

The number and variety of Frankenclouds is sure to increase as new cloud providers emerge and others fall off, and we may bear witness to some pretty strange combinations. The silver lining for IT is the same as it has always been: greater competition drives more choice, product improvements and better prices over time. It may not be pretty, but the Frankencloud is alive -- and that's a beautiful thing.

Alex Barrett is editor in chief of Modern Infrastructure. Write to her at

Dig Deeper on AWS architecture and design

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.