As customized view aggregation is a new feature added to Helix controller, it is hard to predict its impact to the performance of our controller, and the performance of this pipeline is hard to estimate.
We need to support for latency metric reporting between the time when user updates customized state for a partition using CustomizedStateProvider, and the time this change shows up in aggregated customized view and written to zk. With that, we can 1. compare this latency with other pipeline latencies, so we know if this newly added feature saturate our controller or not 2. get an understanding of the pipeline performance 3. we can better monitor this feature and diagnose any problems in the future