One of the interesting consequences of the fragmentation of services (see last post) is the opportunity it presents for the providers of new and innovative services. There has always been a discussion about the challenges of truly achieving economies of scale in IT. Almost all large enterprises with significant IT assets achieve some economies through purchasing power and the sharing of physical assets, but I would assert that few, if any, are able to genuinely attain a state where as they scale up, the overall cost per hypothetical unit of delivery of IT service to the business decreases or even stays constant. In some sense Cloud addresses this and changes it, by turning things on their head. Real economies in IT are perhaps not achieved through scale per se, but rather in combination with specialization. According to most well known cost models, the delivery of IT services is dominated by people and process costs. These tend to be driven by diversity in the IT ecosystem, i.e. by the number of types of services, infrastructure components and so forth, that need to be managed. All of these different things must be coordinated and the business must invest in the people that understand them and the tools that help those people manage them. This becomes highly problematic for large enterprises with huge portfolios of services, sometimes in the tens of thousands, many of which do not deliver differentiating value or are dependent on infrastructure, the management of which does not itself give an advantage.
The trend to Cloud, enabled by the programming model discussed earlier and the evolution of infrastructure from a collection of relatively isolated systems to a virtualized fabric of resources, leads to the opportunity for the optimization and delivery of infrastructure, of platforms and of services by specialist providers. These providers can achieve extreme economies through specialization and scale. By only offering a limited portfolio of services, which are commodities from the perspective of others, but which differentiate the providers, they can minimize diversity and thus management costs. The interesting thing in this world is that the cost model, for the provider, is flipped and becomes dominated by physical infrastructure and power costs. Once they are dominated by these, then economies of scale can be achieved. This is reflected in the drive by many of the classic web 2.0 businesses, i.e. Google, Amazon, eBay, Yahoo et al, to scale using simple, commodity (in essence although sometimes heavily customized) components, and to build scalable and extremely power efficient data centers. Clearly these are not the only companies charging in this direction, but they are leading because their cost model has flipped, and in essence scales based on the number of customers/users they have (which drives infrastructure scale and thus numbers of servers) rather than by the number and complexity of the services they deliver.
Next… The Changing Role of IT Operations