Thomas Watson, a former CEO of IBM, was famously misquoted as saying in the 1940s the world would only need five computers. The misquote is often used as evidence of IBM’s profound misunderstanding of the future potential of computers. I like to think that if Mr. Watson were alive today, he would want this quote attributed to him. The reason: one vision for the future of cloud computing is that the economics of large scale data centers brings an end to the era of the private data center, driving the world to a handful of public cloud providers. As luck would have it, right now there are five likely contenders for the public cloud: Amazon, Microsoft, Google, IBM, and HP.
The alternative vision for the future of course, is that we are headed toward a world of many clouds: a few large public providers, and many private "clouds" in corporate data centers (a.k.a. Virtualization 2.0). Here are my top 5 reasons for not betting on the long term future of private clouds:
#5 - Requires Another Bet On Corporate IT
There has been a simmering dissatisfaction with the service levels and cost-effectiveness of corporate IT since the adoption of the first computers in large enterprises. In IT's defense, they are almost always being asked to do more with less, with highly variable funding and direction from the business from year to year. Over the years, we have had the promise of major technologies transforming corporate IT: time sharing, client-server, web, virtualization, and now cloud. If your company has not realized a substantial transformation of IT in the previous technology shifts, what makes you think you will now? In moments like this, I often reflect on Einstein’s definition of insanity: doing the same thing over and over and expecting different results.
#4 - Pay As You Go Will Win
While the cloud is currently more expensive than physical hardware, over time this gap will diminish through increased competition and innovation among vendors. The combination of reduced cost and higher public cloud security will eliminate the remaining barriers to adoption, enabling early innovators to recognize the full benefits of consumption-based pricing. The benefits of pay as you go will transform the agility of enterprises, providing a competitive advantage that will fuel further adoption.
#3 - Resources Scarcity
Consolidating your applications to highly optimized shared infrastructure requires the development of specialized resources. If you talk to some of the engineers building and supporting public clouds, you will quickly realize that these are not your average corporate IT engineers. Companies like Amazon and Google are assembling some of the smartest technical minds in the industry, producing and operating highly innovative solutions (e.g. data stores with 99.999999999% durability). To compete, corporate IT is going to have to recruit and retain comparable talent, or risk delivering inferior clouds to their businesses.
#2 - Cost Efficiencies of Scale
The greatest cost efficiencies will come from scale, as cloud vendors learn to operate mega data centers with the highest availability, performance and security. Public cloud providers will become increasingly optimized, deriving unique benefits from supporting diverse customers (e.g. the power of non-correlated peaks). In addition, mega data centers will enable lower operational costs, as cloud providers make major capital investments to drive down resource costs and PUE ratios. Some recent examples: Amazon building their own hydroelectric power plant, and Google purchasing wind farms. Competing against public cloud vendors with highly variable IT budgets and project funding will over time be a losing proposition.
#1 - Distraction From Primary Business
The economic environment of the last decade has driven many enterprises to refocus on their strategic core. Complex infrastructure is increasingly becoming available cost effectively from multiple vendors. Just as we would not expect our internal IT department to develop proprietary CRM software in the age of Salesforce, we will not expect them tomorrow to invest in designing, building and operating large scale clouds. With infrastructure as a service, corporate IT will gain the benefits of delegation: managing a vendor to meet the necessary service levels, rather than managing the infrastructure.
I am watching the big movement today toward private "clouds" (a.k.a. Virtualization 2.0). I fully expect this movement to continue for years, possibly even a decade. But at some point, the success of external cloud vendors is inevitable, as innovative CIOs begin to recognize the long term potential of IaaS and PaaS provided by an external vendor. Over time, the concerns about cost and security will dissipate, and the world will coalesce on a few - maybe even five - "computers" in the cloud.
Quote from Amazon presenter at Cloudy Monday's this week: "If your cloud arrives in a box at your loading dock... it's not a cloud."