I think spydum's comment is what you're looking for.

These are things that can and do happen, and you have to plan for it. Designing your applications and infrastructure with eventual consistency in mind can do wonders for both performance and redundancy.