Virtual sprawl is primarily a problem of management – determining and implementing the policies, procedures and workflows necessary to monitor and maintain control over the proliferation of logical assets. The problem comes from the uncontrolled reproduction of virtual devices – it’s like letting all your users or the IT department buy a new physical m/c every time you have a problem while keeping the old one around – pretty soon you just completely overwhelmed with stuff to keep track of. At this point, the task is to structure a management strategy and implementation plan – which should include consideration of future growth. This is a VERY good time to speak with the mainframe staff to pick their brains about how to manage virtualized services and devices.

Virtual stall is primarily a problem of fundamental architecture, i.e. how you will deliver services and how you will structure IT operations – at some point, it appears to be at 30 – 35% infrastructure virtualization – the business managers (hopefully) and the IT staff realize that they are on a path which will fundamentally alter how they operate. Once you past this ‘tipping point’ you fundamentally change how the job of IT. Virtualization is all about providing infrastructure (storage, servers, network access, applications) as a service – it’s the Gateway to the Cloud whether internal, external or hybrid. Somewhere over that percentage your operation becomes indistinguishable from a Cloud (which is simply user transparent access to IT assets they require to get their job done). Think of it as Time-sharing for the 21st century. So, the real decision that faces the manager is how far down the virtualization path they want to go – 99% virtualization can be done on an internal Cloud (can’t be 100% because you do need a physical server/mainframe to run things on). 100% virtualization would be an external Cloud service, and a hybrid is somewhere in-between. Then, they will need a plan on how to get where they want to be.

Finally, a word about mainframe versus server virtualization. Server virtualization is generally motivated by 2 goals: 1) increase utilization (getting above single digit and low teens (15%)) up to 30% or more; and 2) consolidating those sprawling server farms to a few number of physical servers to reduce the management headache associated with an average of 10% (or more) of the physical devices being down at any one time. In fact, it’s trying to get a server to act like a mini-mainframe, high utilization, high reliability, less physical infrastructure to breakdown and a single, integrated, end-to-end management and monitoring view of the environment. Conceptually, the differences are relatively small – so there is a lot that distributed server staff can learn about virtualized environments from mainframes. That said, there are still the implementation differences among the various server virtualization implementations (Citrix, Microsoft, VMware) as well as differences in how the mainframe virtualization is implemented. Both sides would have to learn about the real and faux similarities and the real and faux differences – but that is all part of the dialogue.