IMPROVING THE METHOD OF MATERIALIZATION VIEWS AS A PART OF A DATABASE REENGINEERING PROCESS

Abstract

Data warehouse – is the most important component in the modern cloud web-applications. It has a significant impact on performance of the program system. According to its influence on whole system productivity, incorrect database schemas, none-optimized queries or unsuitable data organization structure can greatly reduce the performance of whole application. With a constantly growing amount of information and brand-new customer requirements that possibly can not be applied to the current architecture, the reengineering process is a strong necessary to refactor and reform the system architecture including data storages. There are a variety of reengineering methods that could make the system faster, easier to maintain and more stable. But some of them are very cost-expensive due to transforming into a more complex formation that it was before, thus maintenance of such system can be not so easy. On the other hand, some of methods are not able to give enough results to fit customer performance requirements. Because of  that, solution should use the known database features with rethinking of its usage as a part of the reengineering process. The potential mechanism of increasing the database performance can be the relational database objects – materialized views with their own refreshing strategies. Having a big read performance, materialized views can be very useful to accessing a large amount of data, but due to some refresh constraints they cannot be used permanently.  So, in the paper the main methods of database reengineering in modern software systems are considered. A comparison and analysis of the most popular ones was performed, and the main stages of data storage reengineering were outlined. The method of materialized views, which is used at the stage of redesigning the tabular structure of the database, has been improved, which allows use the view as one of the optimization components on a permanent basis, thus significantly increases the performance of data selection queries in the software system.