Recent surveys show that nearly all enterprises believe they will adopt a hybrid cloud, but only a very small number...
have actually developed a hybrid cloud strategy, much less deployed one. Because it combines public cloud and private data center principles, it's possible to plan a hybrid cloud from either of these starting points. Picking the better starting point can make your project easier and ultimately more successful. Your goal should always be to minimize change by alternating only what's necessary in your existing infrastructure, and starting in the right place will help you do just that.
To decide where to start a hybrid cloud strategy, first ask, "What is the application relationship across the public-private cloud boundary?" Answering this question establishes the nature of applications' ties to current data center resources. Specifically, do hybrid-exploiting applications elastically expand into the cloud with cloud bursting, do they run in the cloud if the local resources fail using failover, or do some applications run permanently in the public cloud and others run on private resources?
Assessing your existing IT environment
If your applications run in cloud bursting mode, it's almost certain that you'll need to start hybrid cloud planning from the data-center side. Unless you're prepared to rewrite or redeploy existing applications, you'll have to select public cloud services and cloud-bursting practices that are completely compatible with the way your applications currently run.
While failover models of hybrid cloud still benefit from data-center-centric hybrid planning, the all-or-nothing notion of failover may allow you to start by looking for reliable, cost-effective public cloud resources, then seeking an application version that will run on that configuration. An application that runs on one OS in the cloud and another in the data center is a reasonable choice if that configuration is available and cost-optimal.
When some portion of an application will permanently reside in the public cloud, it's smart to plan by starting from the cloud side. Connecting that cloud portion with the rest of the business software set will require integrating across the cloud boundary, and public cloud providers vary in their support and pricing for this kind of connection. By creating an inventory of options stating what public cloud providers can offer, you can then match the choices against the integration options you have available in-house and pick the combination that offers the best capability/cost relationship.
The second question to ask when you're determining where to base your hybrid cloud strategy is, "What is the direction and size of the 'budget flow' across the hybrid boundary?" Generally, a company that's moving most of its IT dollars to the cloud will want to plan that movement based on the public cloud service's cost and features. On the other hand, companies that have relatively small portions of their budgets dedicated to the cloud should emphasize their internal IT plans and architectures in hybrid planning.
Companies that conduct tests to validate the economics of their hybrid cloud strategy -- as all companies should -- should be confident in their project and can give more weight to the public cloud side of their planning. Without proper planning, however, there is a real risk of suspending migration or moving applications entirely back on-premises. In this case, if companies haven't tested for the economics of a hybrid move, they will find that changing on-premises applications to accommodate their public-cloud hosting choice will be a grave error.
Planning a migration based on your cloud bias
If your hybrid planning is biased to the public-cloud side, then your goal in migration is to match the applications to the cloud. Start by determining what your optimum public-cloud choice would be -- based on pricing and features -- then work to adapt your applications' hybridization and integration plans to suit that choice. If your first choice presents too many issues in hybridization, move on to the second choice and so forth, until you get a match.
If the public/private bias lands on the private side, the goal is to create a cloud framework optimized to your existing applications. In most cases, that will start by your examining the interfaces that must be created and sustained across the public/private boundary -- which are defined by the current application workflows -- and then matching cloud capabilities and costs to the support of these specific interfaces. Users report that the pricing model for data access into and out of the cloud is the most critical point, so it's vital that you define these interfaces correctly and price them accurately based on the volume of data.
About the author
Tom Nolle is president of CIMI Corp., a strategic consulting firm specializing in telecommunications and data communications since 1982.