basically, when a request comes in, it picks the node most likely to have the result in its existing cache