When choosing a cluster software stack, most companies undertake a detailed product evaluation phase in which they consider factors such as software license and support costs, maintenance issues, their own use cases, and their existing and planned infrastructure.
However, they often do not pay equal attention to their Grid Engine configuration, which may result in sub-optimal cluster performance and issues like delays for execution of high priority jobs and missing deadlines. These problems are directly related to the company's ability to scale cluster usage: As the user base grows and the user computing needs become more diverse, it becomes more likely that the company would benefit from investing time and resources into properly designing and implementing their cluster configuration.
In this document, Univa UD outlines a service offering for custom-tuning a Grid Engine installation to optimize performance.
SearchStorage.co.UK Research Library
