Performance Testing for Calendars
Harish: The following is an unfinished list of scenarios that need to be tested. Feel free to throw in additional scenarios.
Initial loading of a calendar
Measure the time taken for the initial loading of a calendar.
- Time taken to download the entire calendar (GW - time to create the backend cache)
- Time taken for the current view to be updated with the events that fall within its range, from the time the calendar was enabled by checking it.
- Time taken to display the events after a random jump to a different view/date/month/year
- Time taken to load the calendar component
- Time taken to switch to the calendar component
- Time to drag an appointment (there is a delay when you release it)
- Selecting various ranges of days
Local calendars, GroupWise, Web calendars (On The Web), Exchange
The following are the variants for generating the scenarios.
- Calendar size with respect to the number of calendar items.
- Different views on selection - varying the number of items that might fall in the range of the initial query
- Items with huge and a large number of attachments.
- Calendars with recurrence items - Backends that do not have a recurrence master object, like GW provide a different performance vector
Performance to be measured on
- first time loading after account creation.
- initial loading in subsequent sessions.