This document will help you gain performance when analyzing historical data with PivotalVRP in large GPDB & HAWQ environments.
Issue: When accessing certain graphs PivotalVRP will take a long time to display the information.
Reason: PivotalVRP keeps a raw data repository of every single slice running is each and every segment. Accessing the historical data will sometime cause time-outs and freezes.
Solution: There are several little tricks which can improve the user experience. Freeze the live dashboard while running historical analysis - if you don't need current live information while you conduct a performance drill down, freezing the dashboard will provide a performance boost. Use shorter time periods - When accessing raw data, limiting the time frame to shorter bursts will speed up graph performances, this will be noticeable when using the playback and performance BI modules during heavy load periods. Conduct historical analysis on time frames that start at least a day back - certain data is summarized at the end of each day, in order to gain from the summaries, make sure that your drill down period starts at least a day earlier.