A Network Performance Sensitivity Metric for Parallel Applications

Evans, Janice Perry; Hood, Cynthia S.

doi:10.1007/978-3-540-74742-0_81

Cited by 4 publications

(2 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Assume T idle has exponential distribution (based on the assumption that job submission is a Poisson process). The energy saving can then be written as (6), where λ is the job arrival rate in arrivals per minute. Figure 3 shows the relation between power saving of one cluster node and the product of job arrival rate and the fixed timeout threshold based on the values presented in Table 1.…”

Section: Fixed Timeout Dpm Policymentioning

confidence: 99%

“…While seemingly negligible, any additional levels of traffic between compute nodes can negatively affect the run time of parallel applications, as identified by tools developed in [5] and parallel application run time sensitivity metrics identified in [6]. The tools demonstrate that it is not difficult to render the run time of some parallel applications unpredictable, as the mean run time can be affected by over a factor of two, while the statistical variation of run time can be as high as a factor of 10.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Power and environment aware control of Beowulf clusters

Evans

2009

Cluster Comput

Self Cite

View full text Add to dashboard Cite

Beowulf clusters are now deployed worldwide, chiefly in support of scientific computing. Beowulf clusters yield high computing performance, yet they also pose several challenges: (1) heat-induced hardware failure makes large scale commodity clusters fail quite frequently and (2) cost effectiveness of the Beowulf cluster is challenged by the fact that it lacks means of adapting its power state according to varying work load. This paper addresses these issues by developing a Power and Environment Awareness Module (PEAM) for a Beowulf cluster. The busty nature of computation load in an academic environment inspired the implementation and analysis of a fixed timeout Dynamic Power Management (DPM) policy. Today it is common that many Beowulf clusters in academic environment are composed of older, recycled nodes that may lack of out-ofband management technologies, thus Advanced Configuration and Power Interface (ACPI) and Wake-on-LAN (WOL) technology is exploited to control the power state of cluster nodes. A data center environment monitoring system that uses Wireless Sensor Networks (WSN) technology is developed and deployed to realize environment awareness of the cluster. Our PEAM module has been implemented on our cluster at Purdue University, reducing the operational cost and increasing the reliability of the cluster by reducing heat generation and optimizing workload distribution in an environment aware manner.

show abstract

Section: Fixed Timeout Dpm Policymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%