SOFTWARE PERFORMANCE ENGINEERING AUTOMATION AND TOOL INTEGRATION

SOFTWARE PERFORMANCE ENGINEERING (SPE) TECHNIQUES HAVE THE POTENTIAL TO REDUCE COST AND IMPROVE RELIABILITY OF SYSTEMS. THE TECHNIQUES ALLOW QUANTITATIVE ASSESSMENT OF THE PERFORMANCE CHARACTERISTICS OF EMERGING SOFTWARE SYSTEMS. THIS PERMITS THE SELECTION OF REQUIREMENTS AND DESIGN ALTERNATIVES THAT WILL MEET PERFORMANCE OBJECTIVES PRIOR TO IMPLEMENTATION. SPE HAS BEEN DEMONSTRATED TO BE EFFECTIVE DURING THE DEVELOPMENT OF MANY LARGE SYSTEMS. DESPITE SPE'S SUCCESSES, IT STILL FACES TECHNICAL BARRIERS THAT HINDER ITS WIDESPREAD USE. DEVELOPERS CANNOT EFFECTIVELY EXPLORE DESIGN ALTERNATIVES DUE TO THE LARGE AMOUNT OF MANUAL WORK, THE NEED FOR SPECIALIZED MODELING KNOWLEDGE, AND THE DIFFICULTY IN MAINTAING CONSISTENCY BETWEEN DESIGN MODELS AND SPEMODELS. THIS RESEARCH ADDRESSES THE KEY REQUIREMENTS FOR EFFECTIVE SPE TOOL SUPPORT. IT EVALUATES THREE CANDIDATE ARCHITECTURES FOR SPE SUPPORT THAT INTERFACE WITH BOTH CASE TOOLS AND PERFORMANCE MODELING TOOLS TO CAPITALIZE ON EXISTING TECHNOLOGY. IT IDENTIFIES THE KEY TECHNICAL SOLUTIONS THAT WILL MAKE AN SPE ARCHITECTURE FEASIBLE: A RIGOROUS SPECIFICATION FOR THE INFORMATION REQUIREMENTS TO SUPPORT SPE; A PERFORMANCE META-MODEL THAT DEFINES SPE REQUIREMENTS FOR A VARIETY OF CASE AND PERFORMANCE MODELING TOOLS; AND A PERFORMANCE MODEL INTERCHANGE FORMAT (PMIF) TO EXCHANGE INFORMATION BETWEEN VARIOUS PERFORMANCE MODELING TOOLS.