The current Top SQL report averages usage of queries over the entire period of the request graph. This can lead to some misleading trend information.
Consider the situation where a heavily used Top SQL query has a new version created and an application update switches most of the usage to the new version. The day after the change the 30 day graph scale would make it appear that the overall SQL usage dropped off significantly even if the new query had the same or poorer performance when compared to the old version. This is due to how the averaging is done over the full time period - pulling a single bar day chart for the day before and after the change would be the only way to show the change on the same scale. I've attached some screen caps of this occurring in a real life situation.
To solve this and provide the "at a glance trend" view of Top SQL performance I'd like to suggest an alternative to the Top SQL report is created that uses a consistent scale. This report would only average the information for each bar in the graph individually when displayed so a 30 day chart would be equivalent to what is achieved by pulling 30 individual day Top SQL reports and graphing them together. This would provide a useful trend view of Top SQL resource usage and it's changes day to day. If a major update is made it's relative change to overall SQL usage would be reflected accurately immediately.
The exact time intervals and periods used (day vs month, etc.) could be configurable - the idea is the same.
