The "tactical to infrastructure" evolution nails a common data engineering trap. Making decommissioning criteria explicit from day one could prevent so many platforms becoming archaeological sites
Brilliant framing of how dependency accumlation happens through diffusion rather than deliberate design. The insight about staffing these "simple maintenance" pipelines with junior engineers while seniors move on is especially sharp becasue it creates exactly the knowledge vacuum that makes refactoring risky. The sunset protocol idea is practical, without retirement criteria from inception you really are just building an archaeological site one cron job at a time.
The "tactical to infrastructure" evolution nails a common data engineering trap. Making decommissioning criteria explicit from day one could prevent so many platforms becoming archaeological sites
Brilliant framing of how dependency accumlation happens through diffusion rather than deliberate design. The insight about staffing these "simple maintenance" pipelines with junior engineers while seniors move on is especially sharp becasue it creates exactly the knowledge vacuum that makes refactoring risky. The sunset protocol idea is practical, without retirement criteria from inception you really are just building an archaeological site one cron job at a time.
Yep! Putting juniors on important legacy systems is how temporary becomes permanent. Sunset dates make us call a spade a spade.
Thanks for the shout-out!
Absolutely! It’s a great piece!