Probability, statistics, and queueing theory with computer science applications
Probability, statistics, and queueing theory with computer science applications
Computer Performance Modeling Handbook
Computer Performance Modeling Handbook
| Hi-index | 0.02 |
The performance modeling of computer systems plays an important role in the system engineering life cycle. Historically, however, this important role has not been recognized, and often it has been treated as an obligation. The reason for this diminished stature has many causes, but foremost is that performance modeling has failed to position itself as an indispensible tool to system designers. This single cause itself has many contributors, two of which are the most visible. First, modeling has come to be viewed as an esoteric exercise because of its inability to reflect the system architecture and other system characteristics with ease. Second, the output of most models is a set of statistical data, hardly pointing to a specific design deficiency or operational failure.Two important events in software and language development have come to offer potential solutions to these problems. Object oriented languages allow hierarchical and graphical definition of system architecture, and languages such as PROLOG facilitate the development of knowledge-based systems. We will show how the combination of an object oriented language (Smalltalk) and PROLOG can be used to develop a tool containing hierarchical system description, graphical system entry, performance prediction algorithms, and knowledge-based diagnostic capabilities at an order of magnitude reduction in development costs over standard high order languages such as PASCAL.