| Summary: | Data normalization and denormalization processes are common in database design community as these processes have a great impact on the underlying performance. Current data warehouse queries involve a set of aggregations and joining operations. Thus, normalization process is not a good choice as many relations need to be merged in order to answer queries involving aggregation. On the other hand, denormalization process engages a lot of administrative task. This task takes into account the documentation structure of the denormalization assessments, data validation, schedule of migrating of data and so on. In this paper, we show that the mentioned justifications can not be convincible reasons, under certain circumstances, to ignore the effects of denormalization. Until now denormalization techniques have been introduced for various types of database design. One of the techniques is hierarchical denormalization. Our experimental results indicate that the query response time is significantly decreased when the schema is deployed by hierarchical denormalization on a large dataset with multi-billion records. Thus, we suggest that hierarchical denormalization could be considered as a fundamental method to enhance query processing performance.
|
|---|