ИСТИНА |
Войти в систему Регистрация |
|
ФНКЦ РР |
||
The everyday practice of any large HPC center is full of management and analysis challenges that target altogether the goal of an efficient running HPC center and its systems. This task decomposes into multiple related subtasks: efficient execution of parallel applications and resource utilization, reliability, competent user support, dexterous workflow, etc. Most HPC centers tackle these challenges using diverse software tools that are not originally supposed to be integrated. This entails numerous manual operations and errors in data synchronization. At the same time, many tools meet certain restrictions by vendors and aren’t flexible and portable enough. Tools of our open source set share common workflow and data sources. Data from outer source is used throughout the whole set in a consistent way. It allows avoiding common contradictions in data interpretation when the data is obtained from different unsynchronized sources and helps to automate many steps of the overall working process.