4mbench: Performance Benchmark of Manufacturing Business Database

4mbench: Performance Benchmark of Manufacturing Business Database

연구 분야: Databases

학회: Technology Conference on Performance Evaluation and Benchmarking

A massive number of networked sensors are incorporated into the manufacturing business field, enabling its whole manufacturing process to be monitored and transformed into digital records. Intensive analysis of these digital assets has the potential to offer novel solutions: identifying a problematic piece of the production line and reducing its yield loss. This paper presents 4mbench, a performance benchmark for decision support database in the manufacturing business. 4mbench has employed the 4m (man, machine, material and method) model in order to organize manufacturing event records into relational database and allow business questions to be queried on those event records. This paper presents an overall design of 4mbench that simulates food processing and packaging business operations; specifically, a database schema, a dataset generation rule and a set of queries are introduced. In addition, this paper presents an experimental case study that we conducted with 4mbench on PostgreSQL. This study revealed that the existing query optimization might yield detrimental query execution plans that offered significantly (up to four orders of magnitude) longer execution time. We hope that 4mbench provides researchers and developers with opportunities to explore the scope of further performance optimization on manufacturing business database.