Hacker News Reader

Yes, the main difference would be that distributed systems have more failure modes to be concerned with and that dictates a lot.

In theory it's the same if your model allows for the scheduler to preempt a thread and never schedule it again.