Building a private cloud doesn't have to be an all-or-nothing operation -- there are safe on-ramps for this maze of highways.
Turning back (or at least controlling) the tide of cloud service providers has become the CIO's nightmare. Many businesses are turning to outside providers that can deliver services faster than you can. So the pressure is on to build your on-premises cloud. But how do you get started?
Let's pause for a minute and think about what we are really trying to achieve. After all, you can't simply turn on a cloud, unless you are lucky enough to have a new project, fully funded and well defined, that fits into a "cloud in a box" solution. For large organizations, where is the starting point?
Having been in business for a long time and having had staff turnover probably means that you have well-documented manual processes in place. There are other reasons for documenting processes, including compliance with regulatory requirements or programs such as Six Sigma. Whether these processes are followed or not is always the challenge, but a mature organization will have a fairly high rate of compliance.
These processes can form the basis of the use cases for orchestration and automation. Orchestration is the intelligence behind the automation engine today. Where are your pain points? Where does the business stand to gain the most value from automating a process? There are many options out there for automation software, and you may well feel compelled to leap all in and provide a service catalogue to the business. But hazards beneath the waterline could easily doom you to failure. The on-ramp is to provide an internal, IT-facing catalogue using automation processes that speed your response to the business. Once you have achieved that and have proven the ROI to yourself and to the business, you can build on that foundation.
Let's look at an obvious use case -- one that appears uncomplicated on the surface: automating the provisioning of a virtual machine. Let's look at the resources required for that virtual machine today: CPU, memory, disk, networking, OS, unique identifiers assigned by OS, etc. Each of them could have different requirements for different use cases, but let's assume two or three simplified, common gold standards. Having the resources available just in time wouldn't work today, because you could easily run out of resources. Remember the introduction of auto-extend for databases? I still cringe at the memory. We have to pre-provision these resources, and to do that, we need an idea of how much we're going to need for each budget cycle, at least.
Now on to the next challenge: budgeting. I would assume most organizations have an IT budget that sets aside resources for approved projects, growth, maintenance, etc. If these approved projects are owned individually by departments, you will have to build the business case for pooling the money to buy pooled resources. This again goes back to knowing your pain points, choosing your best opportunity to show real business value, and building from there. Remember the words for success with the business: ROI, ROI, ROI.