Coscheduling under Memory Constraints in a NOW
MetadataShow full item record
Networks of Workstations (NOW) have become important and cost-effective parallel platforms for scientific computations. In practice, a NOW system is heterogeneous and non-dedicated. These two unique factors make scheduling policies on multiprocessor/multicomputer systems unsuitable for NOWs. However, the coscheduling principle is still an important basis for parallel process scheduling in these environments. We propose a new coscheduling algorithm for reducing the number of page faults across a non-dedicated cluster by increasing the execution priority of parallel tasks with lower page fault rate. Our method is based on knowledge of events obtained during execution, as communication activity and memory size of every task. The performance of our proposal has been analyzed and compared with other coscheduling implementations by means of simulation.
Is part ofLecture Notes in Computer Science, 2000, vol. 2221, p. 41-65
European research projects
Showing items related by title, author, creator and subject.
Giné, Francesc; Solsona Tehàs, Francesc; Hernandez, Porfidio; Luque, Emilio (SAGE, 2003)Our research is focused on keeping both local and parallel jobs together in a non-dedicated cluster and scheduling them efficiently. In such a system, memory becomes a critical resource for both kinds of job. Thus, the ...
Giné, Francesc; Solsona Tehàs, Francesc; Hanzich, Mauricio; Hernández Budé, Porfidio; Luque, Emilio (Springer Verlag, 2007)Implicit coscheduling techniques applied to non-dedicated homogeneous Networks Of Workstations (NOWs) have shown they can perform well when many local users compete with a single parallel job. Implicit coscheduling deals ...
Solsona Tehàs, Francesc; Giné, Francesc; Hernández Budé, Porfidio; Luque, Emilio (Iberoamerican Science & Technology Education Consortium, 2001)Our efforts are directed towards the understanding of the coscheduling mechanism in a NOW system when a parallel job is executed jointly with local workloads, balancing parallel performance against the local interactive ...