Performance Evaluation and Prediction for Legacy Information Systems

  • Authors:

  • Yan Jin;Antony Tang;Jun Han;Yan Liu

  • Affiliations:

  • Swinburne University of Technology, Australia;Swinburne University of Technology, Australia;Swinburne University of Technology, Australia;National ICT Australia

  • Venue:
  • ICSE '07 Proceedings of the 29th international conference on Software Engineering
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Database-centric information systems are critical to the operations of large organisations. In particular, they often process a large amount of data with stringent performance requirements. Currently, however, there is a lack of systematic approaches to evaluating and predicting their performance when they are subject to an exorbitant growth of workload. In this paper, we introduce such a systematic approach that combines benchmarking, production system monitoring, and performance modelling (BMM) to address this issue. The approach helps the performance analyst to understand the system's operating environment and quantify its performance characteristics under varying load conditions via monitoring and benchmarking. Based on such realistic measurements, modelling techniques are used to predict the system performance. Our experience of applying BMM to a real-world system demonstrates the capability of BMM in predicting the performance of existing and enhanced software architectures in planning for its capacity growth.