The future of migrations, again
I've written about AI-assisted migrations before (see here and here). But those are about what happens after you've decided to migrate something. What about the decision to do a migration in the first place?
Many migrations don't have to happen immediately or at all. They're opportunities to fix technical debt, try new technology, or prepare for the future. Often, there's no structural reason to do these by migrating rather than by changing things in place. Many migrations happen because they are easier to justify to management, less painful for those whose mistakes caused the problem, and more fun. (I often recall the old sports saying that even when losing isn't the manager's fault, "it's easier to get rid of one manager than 25 players.")
Here as elsewhere, this sounds more cynical than it is. Communicating in organizations is already hard, and justifying fix-it projects is even harder. "Migrating has wonderful benefits and also gets rid of some tech debt" usually sounds better than "hey, we messed up and would like to spend six months fixing our mistakes."
Suppose I'm right (in the posts I link above) that migrations are going to get so much easier than their basic structure will change. What will happen to these "migrations of convenience"? Some possibilities:
- It will be harder to do other work under the aegis of a migration, so that work simply won't get done.
- Migrations will still involve a lot of fiddly, time-intensive work that will create the necessary space (and budget) for the peripheral work.
- Migrations themselves will be a lot faster, but we won't work too hard to communicate that fact to higher-ups. Anecdotally: things like "set up a new database" and "evaluate a broken deployment" routinely draw effort budgets that would have been appropriate c. 2017, and that presents opportunities to do peripheral work. Future migrations might be analogous.
- AI will make us so good at avoiding tech debt that the problem will go away.
There are many other possibilities, and this will play out differently in different settings. (Prediction is hard!) I will be keeping a close eye on the yearly Stack Overflow survey.