Characterizing Machines and Workloads on a Google Cluster
- 1 September 2012
- conference paper
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 397-403
- https://doi.org/10.1109/icppw.2012.57
Abstract
Cloud computing offers high scalability, flexibility and cost-effectiveness to meet emerging computing requirements. Understanding the characteristics of real workloads on a large production cloud cluster benefits not only cloud service providers but also researchers and daily users. This paper studies a large-scale Google cluster usage trace dataset and characterizes how the machines in the cluster are managed and the workloads submitted during a 29-day period behave. We focus on the frequency and pattern of machine maintenance events, job- and task-level workload behavior, and how the overall cluster resources are utilized.Keywords
This publication has 9 references indexed in Scilit:
- Workload characterization and prediction in the cloud: A multiple time series approachPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2012
- Modeling and synthesizing task placement constraints in Google compute clustersPublished by Association for Computing Machinery (ACM) ,2011
- Towards characterizing cloud backend workloadsACM SIGMETRICS Performance Evaluation Review, 2010
- The Datacenter as a ComputerPublished by Springer Science and Business Media LLC ,2009
- A comprehensive model of the supercomputer workloadPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- The elusive goal of workload characterizationACM SIGMETRICS Performance Evaluation Review, 1999
- Generating representative Web workloads for network and server performance evaluationACM SIGMETRICS Performance Evaluation Review, 1998
- Workload characterization: a surveyProceedings of the IEEE, 1993
- Capacity planning for MVS computer systemsACM SIGMETRICS Performance Evaluation Review, 1979