We recently reviewed a research note about Service-Oriented Infrastructure, written by one of my colleagues. The point of SOI is that it’s not SOA. Rather it’s about the type of infrastructure that is required to establish SOA and, more importantly, to be able to determine the effectiveness or ROI for future SOA initiatives. Basically, SOI is a multi-year initiative that standardizes and flattens IT infrastructure. It involves using consistent and repeatable configurations, setups, and processes.
What I find particularly interesting in the discussion swirling around SOI is how it resonates with other arguments about the development of techno-science. In Science on the Run, Bowker—I’m on a bit of a Bowker-binge at the moment—explores the history of Schlumberger. Of great importance to Schlumberger was getting its advanced technology that worked in the lab to function appropriately in the field. Magnetic impedance readings in a hermetically sealed lab are very different from readings taking from a hole drilled in the middle of a Venezuelan jungle. Schlumberger’s trick was to actually turn the wilderness into a lab by controlling process, arrangement, and time. The first step in their projects was to turn some patch of wilderness into a consistent and regular space that could be disciplined. The whole argument reminds me of Foucault and Latour’s pasteurization of France.
SOI seems to be an initiative that fulfills much the same ambition as Schlumberger’s. It is an effort to standardize and flatten infrastructure so that changes and disturbances become manifest and apparent. This approach is in marked difference to the feracious adhocracy that characterizes many IT departments.