Comparison
Golden hammervsNot invented here
Golden hammer
the answer was Kafka again, and the problem this time was a nightly report over four thousand rows.
Reaching for the tool you know rather than the tool that fits, because familiarity is doing the choosing. Everyone does it, and the tell is that the technology is picked before the requirements are stated. It is worth naming carefully in a review, because the counter-argument is often legitimate — a team that operates one technology well may genuinely be better off with an imperfect fit than with an unfamiliar tool nobody can debug at 3am.
Full entry →Not invented here
the team wrote its own scheduler because the existing one did not fit exactly, and now maintains a scheduler.
Rejecting external solutions in favour of building your own, usually justified by a genuine but minor misfit. The build is systematically undercosted because the ongoing ownership — bugs, upgrades, documentation, the person who leaves — is not counted at decision time. The mirror-image trap is real too: adopting a dependency for something that would be forty lines is its own permanent cost.
Full entry →