Now that we are all becoming ‘experts’ about the transformation being seen in the world of computing, we should all be realizing that there is an exciting opportunity in front of us by leveraging the foundational platform innovations that have occurred. “Hybrid computing” is a phrase used by many of you to describe the various different combinations of traditional servers, virtualized servers and the various flavors of the cloud to conduct work. And to add a bit of a second dimension to this phrase, each of these different computing platforms has a number of major vendor-choices, each with a particular set of capacities and specific dependencies. For the server, major choices include HP, DELL, Lenovo and a slew of other Tier-2 vendors like Intel, Supermicro and Quanta. For virtualization we see VMware, Microsoft, Xen and KVM. And the Cloud is dominated by players like Amazon, Softlayer, NTT, Rackspace, VMware, Sungard and CenturyLink.
Now wearing your newly acquired IT business hat (XL, courtesy of your CFO), you should keep in mind that your goal is to find the most suitable platform for each of your applications, and that the definition of “most suitable” changes from application to application and as a function of time. The actual work to be performed doesn’t actually change in this process, it is just the foundational means to deliver that computing that could change. The catalyst for you to consider moving your workloads to other platforms could be for performance or capacity reasons, could be economically driven, or even based upon disaster-recovery or compliance needs.
While many of us have a great understanding about each of the discrete platforms, the ability to move workloads between traditional servers, virtualized servers and the cloud (public and Private) is less understood and yet has already become a critical success factor to managing the costs of doing work and the reliability of IT. As it turns out, moving workloads is conceptually very easy. Simply package up each of your applications as ‘workloads’ (carefully identifying the dependencies that each unit of work has), and then characterize the various platforms on their ability to deliver those dependencies. To put a workload on a particular platform, you simply grab the unit of workload, add to it the specific target platform wrapper, load it on the target platform and press “GO”. In the reverse direction it works in the exact same fashion. Want to move work between Cloud-A and Cloud-B? No problem. Grab the workload, strip off the Cloud-A wrapper, add back in the Cloud-B wrapper and you are all set.
Conceptually it’s about separating the application doing the work from the intended platform to run that work upon. There are a handful of startups, like Silicon Valley-based Rackware and Rightscale from Santa Barbara, that have set their sights on doing this and a few of these companies have become experts at doing so. They make short-work of migrating workloads between public clouds, private clouds, physical servers and virtual instances. Basically any source can be migrated to any destination.
These are really exciting times and those IT pros that fully embrace hybrid computing and workload migration as a function of value and performance will be rewarded over and over again. So while you have spent the last 25 years talking about doing work in ‘the data center’, going forward we should probably start using a term like ‘the data factory’ or other less restrictive phrase that encompasses computing that may also occur outside of our original four walls.