Summary

This chapter covered data warehouses and explained how operational and analytic systems serve very different purposes—one helping with the execution of a business process and the other helping with the measurement of such a business process. The fact that a data warehouse system is also used differently from an OLTP system created opportunities for a new database engine design, which was implemented using the columnstore indexes feature when SQL Server 2012 was originally released.

Because a typical star join query only uses between 10 percent and 15 percent of the columns in a fact table, a columnar-based approach was implemented, which seemed to be more appropriate than the traditional rowstore. Because processing one row at a time also did not seem to be efficient for the large number of rows processed by star join queries, a new batch-processing mode was implemented, too. Finally, following new hardware trends, columnstore indexes, same as with in-memory OLTP...