Fundamentals of Computer Alori
Fundamentals of Computer Alori
| Hi-index | 0.00 |
Historically, workload characterization, and cluster analysis in particular, has been a proven technique when applied to performance evaluation / capacity planning studies. Given the problem of constructing a synthetic workload that represents a production workload, our goal is to use this technique to identify a concise, yet accurate set of work units that will compose the workload. For IMS, these work units are transactions. Yet the selection of transactions must be done with care; for an additional goal must be to identify a concise, yet accurate set of databases that are required by the transactions. This paper will review clustering techniques, and apply them to drive the transaction selection process. An algorithm is also presented that identifies the technique behind database selection. A case study follows that illustrates the implementation of the methodology.