Bulkheads
Isolating resources per-dependency (e.g. separate connection pools) so one failing dependency doesn't starve everything else.
Why we need this / what value this brings
Isolating resources per dependency means one failing dependency can't starve requests to unrelated, healthy ones.
When to use this
Once a single shared resource pool (e.g. one DB connection pool for all endpoints) risks one slow endpoint starving the rest.
How to use or implement this
Use separate connection pools or thread/worker pools per critical dependency, sized to that dependency's actual needs.
Research questions
- Does the backend share one DB connection pool across all endpoints — is there a noisy-neighbor risk?
Empty folder — drop notes, links, and findings here as you research.