Investigation of cluster and cluster queuing system
Abstract
Cluster became main platform as parallel and distributed computing structure
for high performance computing. Following the development of high
performance computer architecture more and more different branches of natural
science benefit fromhuge and efficient computational power. For instance
bio-informatics, climate science, computational physics, computational chemistry,
marine science, etc. Efficient and reliable computing powermay not only
expending demand of existing high performance computing users but also attracting
more and more different users. Efficiency and performance are main
factors on high performance computing. Most of the high performance computer
exists as computer cluster. Computer clustering is the popular and main
stream of high-performance computing. Discover the efficiency of high performance
computing or cluster is very interesting and never enough as it is
really depending on different users. Monitoring and tuning high performance
or cluster facilities are always necessary. This project focuses on high performance
computer monitoring. Comparing queuing status and work load on
different computing nodes on the cluster. As the power consumption is main
issue nowadays, our project will also try to estimate power consumption on
these special sites and also try to support our way of doing estimation.
Description
Master i nettverks- og systemadministrasjon
Publisher
Høgskolen i Oslo. Avdeling for ingeniørutdanningUniversitetet i Oslo