Organizations often fall into new technologies haphazardly. Different groups discover new ways of working that better meet their needs, and they go for it. Eventually, CIOs find themselves with something of a jumble to manage.
Cloud is a particularly extreme example of this, as cloud services can be quite easy to sign onto. There are certainly categories of cloud services that CIOs don't need to control -- things that are opportunistic, department-specific and that don't need to link into core systems or data. As long as the acquiring business people understand and abide by some basic guidelines (mainly having to do with security), let a thousand flowers bloom.
However, when an organization starts to look to cloud for its enterprise IT needs, that's a very different story. It's becoming commonplace to hear CIOs talk about being "cloud first" or having a goal to become "100% cloud." While this often refers to new systems, it must, inevitably, work its way back into the existing environment as well. Decisions about what to move to the cloud will be based on every organization's unique business needs. That said, the following offers an application-centric framework for making those decisions (from my recent Harvard Business Review Analytic Services paper,"The CIO's Guide to Cloud Computing").
Discover: The first step is to find out what you have: for large, complex organizations, this is not an easy thing to do. Even smaller organizations are plagued by renegade applications and servers under desks. But it’s a necessary first step.
Categorize: Once you know what’s out there, sort your applications into the following categories, recommends Bernard Golden, CEO of Navica, a consulting firm focused on cloud computing and DevOps, and the author of Amazon Web Services for Dummies:
- Old/outdated systems of limited value: These should be ruthlessly eliminated. However, it will require CEO authority as well as CIO influence to pry pet applications out of some business users’ hands. CIOs now have some good models to hold up as examples with GE, Motorola Solutions, AstraZeneca, Telefonica and more.
- Necessary but non-differentiating services: Services like email, office productivity, salesforce automation or conventional enterprise applications should migrate to SaaS. This provides businesses with modern innovations as they emerge – for example, new employee productivity features such as social processing; mobile workforce extensions; or increased analytics – and lets CIOs focus their teams’ efforts on things that make a competitive difference. While a simpler choice than rewriting an application, moving to SaaS “is not as simple as you stop using one and start using another,” said an IT leader with over a decade of cloud experience. First you have to be clear with the users of the system that the chosen service will define the process. This may require some negotiation. You have to populate and map the data, and ensure you are able to audit access to it. Finally, the users of the system must be retrained in the new process.
- Necessary and differentiating applications: This is the most challenging group to sort out. Systems that are unique, homegrown or heavily customized are not good candidates for SaaS, but they may be for PaaS or IaaS. Deciding how to handle them requires sorting through many considerations such as the complexity of the code base, portfolio dependencies, network latency issues, whether or not demand is consistent or variable, data residency requirements for the countries in which you operate, and security, privacy and compliance concerns.
In my next post, I'll describe four options for how to approach this last category of systems.
For more on this topic:
Download the PDF of "The CIO's Guide to Cloud Computing"
Watch my presentation at CIO Sweden's Cloud Confessions Conference