Productivity may not be a simple function of person-hours, but having zero person-hours available because the persons are preoccupied with legacy services will kill a launch or landing. Turning legacy off is important for sub-linear headcount growth.
The most promising targets for turndown are:
- Old systems riddled with technical debt;
- Internal tools and services nobody cares for anymore;
- Yesterday's formerly-shiny ${THING} that ${THING2} has replaced;
- Services at least one or two layers beneath what really matters to the organisation.
But there are always blockers even when the goal is clear and the motives are strong:
- The system is in the critical path for something critical;
- Other systems rely on it in some long-forgotten way;
- The organisation's current "Death Star megaproject" is using it and you can't shift their migration timeline;
- It offers some useful feature that the replacement doesn't have.
In this talk I will explain why turning things off (for good) is desirable, describe a few services my team was responsible for turning down, how they related to other services with illustrations, what made turning down possible, and how we got there in the end (or didn't get there).