Generally speaking, you create summaries of data to report on big data. These summaries are created by background runs of the search on which your report is based. When a user runs a report against data that has been pre-summarized, it runs considerably faster because the summaries it uses are much smaller bits of the total number of events to be searched.
This concept of pre-summarizing events can be used with your reports, data models, and indexes. Right now, we will focus on report summary acceleration.