Improving Speculative Execution Performance with Coworker for Cloud Computing

Huang, Sheng Wen; Huang, Tzu‐Chi; Lyu, Syue-Ru; Shieh, Ce-Kuen; Chou, Yi-Sheng

doi:10.1109/icpads.2011.72

Cited by 6 publications

(8 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These approaches primarily use speculative execution based methods that create replicas of detected stragglers which leverage redundant computation [7][8], network congestion [9], and data locality [10] to reduce overall job completion time. While such works have been demonstrated to reduce the impact of stragglers upon service operation, their effectiveness is dependent on realistic assumptions pertaining to system behavior.…”

Section: Introductionmentioning

confidence: 99%

Straggler Root-Cause and Impact Analysis for Massive-scale Virtualized Cloud Datacenters

Garraghan

Ouyang

Yang

et al. 2019

IEEE Trans. Serv. Comput.

View full text Add to dashboard Cite

Abstract-Increased complexity and scale of virtualized distributed systems has resulted in the manifestation of emergent phenomena substantially affecting overall system performance. This phenomena is known as "Long Tail", whereby a small proportion of task stragglers significantly impede job completion time. While work focuses on straggler detection and mitigation, there is limited work that empirically studies straggler root-cause and quantifies its impact upon system operation. Such analysis is critical to ascertain in-depth knowledge of straggler occurrence for focusing developmental and research efforts towards solving the Long Tail challenge. This paper provides an empirical analysis of straggler root-cause within virtualized Cloud datacenters; we analyze two large-scale production systems to quantify the frequency and impact stragglers impose, and propose a method for conducting root-cause analysis. Results demonstrate approximately 5% of task stragglers impact 50% of total jobs for batch processes, and 53% of stragglers occur due to high server resource utilization. We leverage these findings to propose a method for extreme straggler detection through a combination of offline execution patterns modeling and online analytic agents to monitor tasks at runtime. Experiments show the approach is capable of detecting stragglers less than 11% into their execution lifecycle with 95% accuracy for short duration jobs.

show abstract

Section: Introductionmentioning

confidence: 99%

Straggler Root-Cause and Impact Analysis for Massive-scale Virtualized Cloud Datacenters

Garraghan

Ouyang

Yang

et al. 2019

IEEE Trans. Serv. Comput.

View full text Add to dashboard Cite

show abstract

“…For Reduce skew handling approaches, Co-worker [10] functions in a way that as long as a straggler is identified, the reserved co-worker task will help process the remaining data. Its effectiveness is dependent on the choice of the reserved coworker number, and will introduce resource overhead when there is no skew.…”

Section: Related Workmentioning

confidence: 99%

Mitigate data skew caused stragglers through ImKP partition in MapReduce

Ouyang

Zhou

Clement

et al. 2017

2017 IEEE 36th International Performance Computing and Communications Conference (IPCCC)

View full text Add to dashboard Cite

Speculative execution is the mechanism adopted by current MapReduce framework when dealing with the straggler problem, and it functions through creating redundant copies for identified stragglers. The result of the quicker task will be adopted to improve the overall job execution performance. Although proved to be effective for contention caused stragglers, speculative execution can easily meet its bottleneck when mitigating data skew caused stragglers due to its replication nature: the identical unbalanced input data will lead to a slow speculative task. The Map inputs are typically even in size according to the HDFS block configuration, therefore the skew caused stragglers happen mainly in the Reduce phase because of the unknown intermediate key distribution. In this paper, we focus on mitigating data skew caused Reduce stragglers, propose ImKP, an Intermediate Key Pre-processing framework that enables the even distributed partition for Reduce inputs. A group based ranking technique has been developed that dramatically decreases the pre-processing time, and ImKP manages to eliminate this timing overhead through parallelizing the pre-processing with the file uploading procedure (from local file system to HDFS). For jobs that take input directly from HDFS, ImKP minimizes the overhead by storing the < GroupedKey, Reducer > mapping result on every node within the cluster for reuse. Experiments are conducted on different datasets with various workloads. Results show that, compared to the popular hash partition, ImKP can dramatically decrease Reduce skew, achieving a 99.8% reduction in the coefficient of variation of the input sizes in average, and improve up to 29.37% job response performance.

show abstract

“…1). The later-completing jobs that comprise this heavy tail are sometimes referred to as "stragglers" [18,19]. By standardizing the length of GARLI workunits, we aim to improve overall analysis batch turnaround timei.e., reduce the straggler effect -by decreasing the variance in analysis runtimes.…”

Section: Problem Description and Proposed Solutionmentioning

confidence: 99%

Subdividing Long-Running, Variable-Length Analyses Into Short, Fixed-Length BOINC Workunits

Bazinet

Cummings

2015

J Grid Computing

View full text Add to dashboard Cite

We describe a scheme for subdividing longrunning, variable-length analyses into short, fixedlength BOINC workunits using phylogenetic analyses as an example. Fixed-length workunits decrease variance in analysis runtime, improve overall system throughput, and make BOINC a more useful resource for analyses that require a relatively fast turnaround time, such as the phylogenetic analyses submitted by users of the GARLI web service at molecularevolution.org. Additionally, we explain why these changes will benefit volunteers who contribute their processing power to BOINC projects, such as the Lattice BOINC Project (boinc.umiacs.umd.edu). Our results, which demonstrate the advantages of relatively short workunits, should be of general interest to anyone who develops and deploys an application on the BOINC platform.

show abstract

Improving Speculative Execution Performance with Coworker for Cloud Computing

Cited by 6 publications

References 12 publications

Straggler Root-Cause and Impact Analysis for Massive-scale Virtualized Cloud Datacenters

Straggler Root-Cause and Impact Analysis for Massive-scale Virtualized Cloud Datacenters

Mitigate data skew caused stragglers through ImKP partition in MapReduce

Subdividing Long-Running, Variable-Length Analyses Into Short, Fixed-Length BOINC Workunits

Contact Info

Product

Resources

About