LEDE: One of the hardest parts of DevOps movement is explaining the unique value to IT Leadership in conventional organisations that rely on ITIL principles. I’m having success by framing the debate in terms of over-capitalised on assets and under-invested in human infrastructure.
Three Infrastructure Types
There are three types of infrastructure in a corporate IT setting – Physical, Virtual and Human.
Physical Infrastructure – the physical assets like the data centre, server, routers and switches. Well known today and provided by vendors who fund large and extensive training programs on their products (also called marketing).
Virtual Infrastructure – the software and applications that run over the top of the physical infrastructure, also well funded from product sales as a marketing strategy.
Human Infrastructure – this infrastructure is largely funded by the customer and provided internally. Occasional purchases are made by purchasing consultants on the open market or vendor-aligned human products.
The available funds for marketing and sales of Physical and Virtual products has created a market distortion with over-emphasis on repeatedly purchasing new products.
Metaphor: I use this metaphor. When you purchase a car, you expect to take it for maintenance every year or so. After a few years, the car may be replaced for a number of reasons. An “Enterprise IT car” would be replaced every year to ensure that we don’t pay for mechanics in the garage to service it and maintain the vehicle in usable and reliable state.
Increasing Utilisation Will Increase the Rate of Return
I take the view that the heart of DevOps movement is a return to using Human Infrastructure to maximise the return on invested capital. I read somewhere that an average Enterprise IT infrastructure runs at less 5% utilisation.
Network chassis have large numbers of unused slots, backbone bandwidth often runs at less 10% sustained. Storage arrays have unused shelves or reserved capacity for “future growth”. Most server run at less than 5% CPU and memory is under pressure for a few minutes per day.
IT leadership could reduce CapEx by investing in ways to increase utilisation. How can we do this ? In short, by increasing investment in human infrastructure.
DevOps is Human Infrastructure
Like any financial system, IT has developed a viral response to over allocation of capital and poverty in human infrastructure. It’s called DevOps. From my perspective, DevOps is a strategy for increasing investment in operational processes and associated human infrastructure to drive increase in utilisation of physical and virtual assets.
To put that more plainly, hire more people with skills and expertise in automation and orchestration. Harness these skills to increase the utilisation of existing assets and delay, or prevent, capital expenditure. This increase in OpEx and reduction in CapEx is sometimes also known as cloud (whether public or private).
I often simplify human infrastructure to three key values:
The first two are readily available from existing training programs and pedagogically comprehensive systems exist to get them.
But expertise is much harder to build. In real terms, a manager of human infrastructure needs to build factory that produces expertise. But that will be the subject of another blog.