I counted the estate properly last year for the first time. Two hundred and thirty applications for an organisation of about four thousand people. Around sixty of them had no identifiable business owner. Nineteen had not been logged into by a human in over a year but were still running, still patched, still backed up, still costing money and still holding personal data we were obliged to protect.
Nothing about that is exceptional. It is what happens when an organisation is good at starting things for three decades and has never built a competent way to end them. Every one of those systems was justified when it was bought. Every one has a small group of people who would be inconvenienced by its removal, and no group at all whose job is to remove it. Adding a system has a sponsor, a budget line and a launch. Removing one has a risk register and no applause.
The reason I think this becomes the defining problem rather than a housekeeping annoyance is that the rate of addition is about to increase sharply. It is now cheap to produce a working application, and it is going to get cheaper. What has not changed at all is the cost of operating one: patching, identity, backup, exit, data obligations, the person who knows how it works leaving. If creation gets ten times cheaper and operation does not, the ratio of estate to capacity gets worse every year, and it is already bad.
The organisations I expect to do well are not the ones adopting fastest. They are the ones that treat decommissioning as a funded, scheduled activity with a named owner, the same way they treat delivery. That means recording an exit plan at purchase, reviewing systems on a cycle rather than on rumour, and accepting that a certain amount of engineering time each quarter produces no new capability and simply reduces surface area.
The skill I would invest in, if I were starting again, is not another platform. It is being the person who can look at a sprawling estate, work out what is genuinely load-bearing and what is merely still switched on, and safely turn off the rest. That work is unglamorous, permanently in demand, and almost nobody is doing it.
– Serguey Shinder
Top comments (0)