Software Reliability Model for Distributed Systems

Distributed computing has the potential of much further growth aided by current developments in the computer and communication areas, but this progress may be impeded by deficiencies in software reliability and maintainability. The software reliability modeling methodology and tool to be developed under this effort can make outstanding contributions to reducing system and software reliability problems by (a) identifying areas where deficiencies are most likely to be found; these can be functional areas (e.g., redundancy management) or software technology areas (e.g., task synchronization), (b) permitting rapid evaluation of the reliability benefits of alternative software or system configurations, (c) providing an objective basis for hardware/software trade-offs, and (d) providing an objective basis for reliability/performance trade-offs. The effort consists of (1) definition of software reliability metrics for distributed Systems, (2) establishing model selection criteria, (3) model evaluation, and (4) prototyping the selected model.