Nice idea, would this be extended to support other systems beside redis? many of the faults related to network conditions are generic and not related to wire application protocol.
Redis is just the start.
There will be more FIP for TCP in general. And more specific rules for certain protocols atop of it, like HTTP, C* and others.