Avoiding Fallback in Distributed Systems