Consider system that is processing a stream of tuples for a relation r with attributes (A, B, C, timestamp) Suppose the goal of a parallel stream processing system is to compute the number of tuples for each A value in each 5 minute window (based on the timestamp of the tuple). What would be the topic and the topic partitions? Explain why.