Search

The article offers information on how professional software developers can improve their performance standards with effective implementation of test-driven development (TDD) process. It involves writing a unit test followed by writing a code which should fail in the test. Then a new code is written which should pass the test and thus a new capability is added to the software. The author suggests there are three laws to be kept in mind while using TDD which include not writing a code before first writing a failing unit test. Another thing is to not write more of a code than is sufficient to fail in the test and last but not the least not writing more of a code than is sufficient to pass the test.

Custom software development projects under product specification uncertainty can be subject to wide variations in performance (duration, cost, and quality) depending on how the uncertainty is resolved. A prototyping strategy has been chosen in many cases to mitigate specification risk and improve performance. The study reported here sought to illuminate the case where duration was the highest priority project constraint, a feature that has often called for a concurrent development process. Unlike concurrent engineering, however, the process in this paper was a sequential, three-phase approach including an optional, up-front prototyping phase, a nominal-duration construction phase, and a variable length rework phase that grew with the arrival of specification modifications. The source of uncertainty was the modification arrival time; the management control point was the amount of time spent engaged in prototyping activities. Results showed that in situations where the modification arrival rate was sufficiently faster during prototyping than during construction, a minimal-duration choice was available. The model also returned the solution to perform no prototyping in cases where arrival rates were nearly equivalent between phases, or when the rework cost associated with modifications was consistently low. [ABSTRACT FROM AUTHOR]

Management of performance requirements is a major challenge for information systems as well as other software systems. This is because performance requirements (e.g., "a student records system should have good response time for registering students") can have a global impact on the target system. In addition, there are interactions and trade-offs among performance requirements, other nonfunctional requirements (NFRs, or software quality attributes, e.g., accuracy), and the numerous alternatives for the target system. To provide a systematic approach to managing performance requirements, this paper presents a "Performance Requirements Framework" (PeRF) It integrates and catalogues a variety of kinds of knowledge of information systems and performance These include performance concepts, Software Performance Engineering (SPE) principles for "building performance into" systems, and information systems development knowledge (including requirements, design, implementation, and performance). In addition, layered structures organize performance knowledge and the development process. All this knowledge is represented using an existing goal-oriented approach, the "NFR Framework," which offers a developer-directed graphical treatment for stating NFRs, analyzing and interrelating them, and determining the impact of decisions upon NFRs. This approach allows customized solutions to be built, taking into account the characteristics of the particular domain. The use of PeRF in managing performance requirements is illustrated in a study of performance requirements and other NFRs for a university's student records system. This paper concludes with a summary of other studies of information systems, tool support, and directions for future work. [ABSTRACT FROM AUTHOR]

An approach to software performance testing is discussed. A case study describing the experience of using this approach for testing the performance of a system used as a gateway in a large industrial client/server transaction processing application is presented. [ABSTRACT FROM AUTHOR]

The article presents the closely related issues of performance requirements and of testing. Brian Nixon's paper, "Managing Performance Requirements for Information Systems," describes a framework called Performance Requirements Framework (PeRF) for presenting and managing the performance requirements for information systems. Since information system have to maintain a persistent database and provide an interface with external users, with functional and nonfunctional requirements have to be considered. Functional requirements describe what an information system should do, whereas nonfunctional requirements consider software quality factors such as security and reliability. The paper proposes the treatment of performance requirements as non functional requirements and applies PeRF to manage the performance requirements of a University's student records system. Elaine Weyuker and Filippos Vokolos provide a healthy dose of practicality in "Experience With Performance Testing of Software System."