Write distributed systems that are resilient and self-heal. Remote calls can fail or hang indefinietly without a response. Supervision will help to isolate failure and keep individual components from bringing down the whole system.