For Excel PowerPivot, the database is IN the workbook
Concerning Excel PowerPivot, the question is often asked how is it possible for Excel to now handle tens or hundreds of millions of rows – the way native Excel handles tens or hundreds of thousands of rows.
How you ask?
Well, the reason why is because Excel (with the PowerPivot add-in) is using the new In-Memory BI (IMBI) database. It utilizes a column-based store which has been seen academically and empirically to have some interesting characteristics within BI scenarios…namely, it can calculate many (but not all) BI aggregate very quickly. Because it is in-memory, it does not utilize disk I/O and performs all of its scans in-memory – i.e. much faster because you do not have disk contention. As well, it has a great compression algorithm in order to significantly reduce the size of the overall data stored. By the way, this new database architecture is called VertiPaq for all those interested in the trivia!