Extract what hurts, keep what works

When a monolith gets slow or fragile, the big rewrite is tempting. The alternative is to move out only the capabilities that have a specific reason to leave:

• performance: it’s a bottleneck;
• reliability: it needs to be more dependable than it can be where it is;
• independent deployment: it needs its own release schedule.

Everything without such a reason stays where the domain logic already lives.

Where I did this: Moving parts of a Rails monolith to Go, not all of it (Rails → Go, at JobReady).

Open questions

• How do you notice when “selected capabilities” is slowly turning into a rewrite after all?
• What’s the smallest useful seam for the first extraction?