The applicability of factory-floor techniques to IT

In my review of The Phoenix Project I questioned the applicability of factory-floor techniques to IT. There is potential to be exploited there - hence I recommended everyone read the book - but we can go too far with the application of Lean, TQM, Six Sigma etc to IT. Manufacturing techniques are only partly applicable.

IT is different, even though the Phoenix Project does not accept that assertion. When someone of the calibre of Charles Betz says "IT solution development and service delivery are more variable than stereotypical manufacturing processes" I listen (see Charles' brilliant book Architecture & Patterns for IT, pages 9 and 15-16).

ImageI do think the two views can be reconciled, in Standard+Case. S+C is my new approach for IT service support (or any situational response). It is based on a dichotomy, a two-sided model, a Yin and Yang. One side of the S+C approach is Standard: the predefined processes and model workflows which work in simple known situations. The other side is Case: managing unique unpredictable unfamiliar situations which require an expert, or darkly unknown alien situations which require explorers and inventors, or chaotic uncontrolled disasters which require emergency response.

Standard situations are amenable to the application of manufacturing principles. As the name implies, the activities are standardised: predefined, repeatable, measurable. They can be turned into a production line. People can be trained to do them. Factory work. Lean, TQM, and Six Sigma can be used to refine the procedures, prune the value chains, remove variation. Outsource-able. Offshore-able.

Case situations on the other hand refuse to be standardised. Each one unfolds differently. The workflow is emergent. Knowledge workers have to work each case out. There is no substitute for expertise and capability. You could sub-contract it to a partner you trust but you wouldn't hand it off to anyone else. And it is not a factory: it is an artisan's craft workshop; an engineer's laboratory. Repeatability is our enemy: diversity allows us to explore the best ways to deal with a new type of situation, in the hope that if it happens often enough we can start to standardise the response for the future.

Taking this model, we see that development and operations can fulfil the dreams of DevOps and become automated only if it is a Standard application: one that is simple, familiar. Want a website built? Ah yes we know how to do that. We can build a machine to do it: we can crank out a build in minutes. CASE (Computer Aided Software Engineering) come true. But CASE is not case, it is the opposite. CASE only works in situations so Standard that they can be automated. Customising a package on a different database and hardware to create a new core system for a trucking company to allocate routes and loads, and calculate billing by tonnage and distance? Not so much. A new operating system for a dedicated bio-metric device? Also unfamiliar. Which is not to say that we couldn't still construct a machine to crank the code builds out at the push of a button, and to automate the underlying infrastructure as code, but the effort to make the whole of development and operations sufficiently familiar, known, standardised, and repeatable can eat most or all of the savings in cost or time from the ultimate automation. I've said it before: there are times when automation pays and times when it doesn't.

Likewise IT support. About half the incidents and requests that come in are familiar. We can script the response. We can gather statistics; we can improve average performance and reduce variance. The other half are the outcome of a changing world: users want something new; a service has changed; something is running out or broken; the external environment has changed. We can't predict; we can only be well equipped. Lean and Six Sigma techniques will only break Case Management. They squeeze out the slack we need to respond to the unexpected. They destroy the freedom to explore new techniques. They stop the experts from playing to their own individual strengths.

There is a place for both ways of dealing with the world, hence Standard+Case. Together Standard and Case are the Yin and Yang of response. Combined they make a complete approach for dealing with all forms of situations. The manufacturing techniques are only half of the answer.

Syndicate content