From a financial point of view, there is a critical assumption built into the pay-as-you-go model: that consumption and therefore cost is predicable. In the early days public cloud was always cheaper, however, great strides in private cloud technology, maturation of the space, and the challenges in moving to a cloud centric IT platform have muddied the waters. As someone who has never been a proponent of private cloud, I feel the water is as clear as ever. IT infrastructure has never been built on predictability. For decades, networks, servers, and storage have been designed to a “just in case” standard. The result is tremendous bloat which is being engineered out through virtualization, but the workloads are no more or less predictable. What’s missing is the understanding that public cloud isn’t just about capex vs opex, it’s also about momentum and friction.
Infrastructure assets are evil; they anchor us in the technology of the day and act as a damper on innovation. I’ve lived the nightmare of being forced to architect a solution around existing infrastructure assets, sometimes merely to justify their existence regardless of their impact on the solution. Assets grow like a planet, swirling gasses of expectation coming together forming a gravitational field which attract more mass in the form of processes and protections. Soon the planet becomes so massive it starts attracting its own moons of ancillary assets into orbit. Where there’s mass we have to respect Newton’s three laws. Objects at rest tend to stay at rest and thus have no momentum, not the message the CEO wants to hear in relation to innovation. Objects require a force to accelerate dampened by friction; the greater the mass, the more friction will work against the building momentum. And since for every action there is an equal and opposite reaction, to generate the momentum required for change, ever increasing investments in time and energy are required.
Like the car travelling toward the horizon will get closer to the mountains but not the moon, private cloud gets a company into virtualization but not cloud computing. I know some will argue I’m too optimistic, however I point to the success of companies such as Netflix, whose competitive advantage is the result of investing early in learning the lessons of public cloud. I also point to respected executives like Chris Drumgoole, COO of IT at GE, making bold public statements about their migration not just to cloud, but to public cloud. What he has said publicly many others have told me privately, but first they must get their CFO’s to look past the head fake of relating predictability to public cloud risk.
