Abstract:

A data warehouse stores historical data in order to support decision making. This data, reflecting the trends in data over time, is used for analytical query processing. The analytical queries are usually long and complex and processing them on a large data warehouse usually consumes a lot of time. As a result, the query response time is high. This high query response time needs to be reduced in order to make analytical query processing more efficient. One way to do this is by using materialized views. Materialized views contain pre-computed and summarized information stored in a data warehouse for the purpose of answering analytical queries. This necessitates that these materialized views contain relevant and required information for processing analytical queries. As the number of possible views is exponential with respect to the number of dimensions, materializing all views is not feasible due to space constraints. Therefore, there is a need to select a subset of views, from among all the possible views, for materialization. Selecting optimal subsets of views is an NP-Complete problem. Alternatively, views can be selected based on heuristics like greedy, evolutionary etc. This paper focuses on the greedy based selection of materialized views wherein, at each step, the most beneficial view is selected for materialization. Most greedy based approaches do not take into consideration the views that are frequently accessed by queries posed in the past. These views, referred to as frequent views, have high likelihood of being accessed again in future and are therefore good candidates for materialization. The approach presented in this paper identifies such frequent views, from among all views in the lattice, and then greedily selects beneficial views from among these frequent views. These selected views are likely to improve the query response time and thereby facilitate decision making. Read More

Meet one of the Author:

Affiliations: