Adaptive scheduling with parallelism feedback
Proceedings of the eleventh ACM SIGPLAN symposium on Principles and practice of parallel programming
Adaptive work stealing with parallelism feedback
Proceedings of the 12th ACM SIGPLAN symposium on Principles and practice of parallel programming
Adaptive work-stealing with parallelism feedback
ACM Transactions on Computer Systems (TOCS)
Improved results for scheduling batched parallel jobs by using a generalized analysis framework
Journal of Parallel and Distributed Computing
Provably efficient two-level adaptive scheduling
JSSPP'06 Proceedings of the 12th international conference on Job scheduling strategies for parallel processing
Optimal scheduling of heavy tailed traffic via shape parameter estimation
MILCOM'06 Proceedings of the 2006 IEEE conference on Military communications
Server Scheduling to Balance Priorities, Fairness, and Average Quality of Service
SIAM Journal on Computing
Optimizing the stretch of independent tasks on a cluster: From sequential tasks to moldable tasks
Journal of Parallel and Distributed Computing
Adaptive parallelism for web search
Proceedings of the 8th ACM European Conference on Computer Systems
Hi-index | 0.00 |
We consider the problem of scheduling dynamically arriving jobs in a non-clairvoyant setting, that is, when the size of a job in remains unknown until the job finishes execution. Our focus is on minimizing the mean slowdown, where the slowdown (also known as stretch) of a job is defined as the ratio of the flow time to the size of the job. We use resource augmentation in terms of allowing a faster processor to the online algorithm to make up for its lack of knowledge of job sizes. Our main result is that the Shortest Elapsed Time First (SETF) algorithm, a close variant of which is used in the Windows NT and Unix operating system scheduling policies, is a $(1+\epsilon)$-speed, $O((1/\epsilon)^5 \log^2 B)$-competitive algorithm for minimizing mean slowdown non-clairvoyantly, when $B$ is the ratio between the largest and smallest job sizes. In a sense, this provides a theoretical justification of the effectiveness of an algorithm widely used in practice. On the other hand, we also show that any $O(1)$-speed algorithm, deterministic or randomized, is $\Omega(\min(n,\log B))$-competitive. The motivation for resource augmentation is supported by an $\Omega(\min(n,B))$ lower bound on the competitive ratio without any speedup. For the static case, i.e., when all jobs arrive at time 0, we show that SETF is $O(\log{B})$ competitive without any resource augmentation and also give a matching $\Omega(\log{B})$ lower bound on the competitiveness.