Software Support for Metrology - Good Practice Guide No.17. Distributed computing for metrology applications.

This guide aims to facilitate the effective use of distributed computing methods and techniques by other National Measurement System (NMS) programmes and by metrologists in general. It focuses on the needs of those developing applications for distributed computing systems, and on PC desktop grids in particular, and seeks to ensure that application developers are given enough knowledge of system issues to be able to appreciate what is needed for the optimum performance of their own applications. Within metrology, the use of more comprehensive and realistic mathematical models that require extensive computing resources for their solution is increasing. The motivation for the guide is that in several areas of metrology the computational requirements of such models are so demanding that there is a strong requirement for distributed processing using parallel computing on PC networks. Those who need to use such technology can benefit from guidance on how best to structure the models and software to allow the effective use of distributed computing. This work aims to ensure that metrologists can, when appropriate, take maximum advantage of the significant improvements in computational speed that are offered by cost-effective distributed computing. In addition, it contains case studies and examples that demonstrate what can be achieved with distributed computing technologies.