Enhanced Big Data Analytical Reporting Using Hybrid Materialized Views Storage For Manufacturing Execution System (MES)

Emran, Nurul Akmar and Ahmad, Norashikin and Shaaban, Azizah and Ibrahim, Hamidah and Mohd Yusof, Mokhtar and Md. Bohari, Nor Mas Aina (2015) Enhanced Big Data Analytical Reporting Using Hybrid Materialized Views Storage For Manufacturing Execution System (MES). Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Enhanced Big Data Analytical Reporting Using Hybrid Materialized Views Storage for Manufacturing Execution System (MES).pdf - Submitted Version Restricted to Registered users only Download (984kB)

Abstract

Within organizations that rely on big data to produce analytical reporting, time is a key factor that determines organization’s competitive advantage. In manufacturing industry, data volumes are usually big enough for the analytical reporting to become an issue. In this domain, Manufacturing Execution System (MES) is commonly used to produce analytical reporting in real-time. The speed of reports delivery is determined by the speed of queries executed by the system. MES unfortunately cannot tolerate with delayed queries (and therefore delayed reports) as manufacturing products’ quality and profits can be negatively affected with the delays. Therefore, the need to maintain an acceptable level of real-time reporting in this domain is crucial. Database Administrators (DBAs) use materialized views to speed up the queries. This method is designed for applications that have frequently-executed expensive queries (intensive aggregation and join operations) where the results are stored on disk. However, using materialized view alone is insufficient in dealing with big data. Even though the benefits of materialized views are known in speeding up queries, as shown in this research, there are cases where the classic materialized views are inadequate in guaranteeing acceptable query performance. Thus, this research presents the results of experimenting SQL query rewriting by utilizing ‘multitier’ or hybrid materialized views structure, a new approach that is proposed to extend the classic materialized view’s functionality. In particular, sub-materialized views (SMVs) concept is defined and implemented using real data sets from SilTerra (a semiconductor industry). The outcome of the experiment supports the hypothesis of the research that SQL query rewriting using SMV outperforms the classic rewriting. The results also reveal that, the performance of SMV is far better (than without SMV) for complex queries against large data sets. Even though SMV has been tested using the samples of semiconductor’s data set, other applications with similar problems can benefit from the implementation of it. The research presented in this research contributes towards understanding multitier materialized views concept, the forms and practical aspect of SMV for SQL query rewriting.

Actions (login required)

View Item