New Dimension Value Introduction for In-Memory What-If Analysis
Gaurav Saxena, Ruchi Narula, Manish Mishra · arXiv (Cornell University) · 2013
OLAP systems operate on historical data and provide answers to analysts queries. Recent in-memory implementations provide significant performance improvement for real time ad-hoc analysis. Philosophy and techniques of what-if analysis on data warehouse and in-memory data store based OLAP systems have been covered in great detail before but exploration of new dimension value (attribute) introduction has been limited in the context of what-if analysis. We extend the approach of Andrey Balmin et al of using select modify operator on data graph to introduce new values for dimensions and measures in a read-only in-memory data store as scenarios. Our system constructs scenarios without materializing the rows and stores the row information as queries. The rows associated with the scenarios are constructed as and when required by an ad-hoc query.