On the latency-accuracy tradeoff in approximate MapReduce jobs

Juan F. Perez, Robert Birke, Lydia Y. Chen

Research output: Chapter in Book/ReportConference contribution

4 Scopus citations


To ensure the scalability of big data analytics, approximate MapReduce platforms emerge to explicitly trade off accuracy for latency. A key step to determine optimal approximation levels is to capture the latency of big data jobs, which is long deemed challenging due to the complex dependency among data inputs and map/reduce tasks. In this paper, we use matrix analytic methods to derive stochastic models that can predict a wide spectrum of latency metrics, e.g., average, tails, and distributions, for approximate MapReduce jobs that are subject to strategies of input sampling and task dropping. In addition to capturing the dependency among waves of map/reduce tasks, our models incorporate two job scheduling policies, namely, exclusive and overlapping, and two task dropping strategies, namely, early and straggler, enabling us to realistically evaluate the potential performance gains of approximate computing. Our numerical analysis shows that the proposed models can guide big data platforms to determine the optimal approximation strategies and degrees of approximation.

Translated title of the contributionSobre la compensación de latencia y precisión en los trabajos aproximados de MapReduce
Original languageEnglish (US)
Title of host publicationINFOCOM 2017 - IEEE Conference on Computer Communications
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781509053360
StatePublished - Oct 2 2017
Event2017 IEEE Conference on Computer Communications, INFOCOM 2017 - Atlanta, United States
Duration: May 1 2017May 4 2017


Conference2017 IEEE Conference on Computer Communications, INFOCOM 2017
Country/TerritoryUnited States

All Science Journal Classification (ASJC) codes

  • General Computer Science
  • Electrical and Electronic Engineering


Dive into the research topics of 'On the latency-accuracy tradeoff in approximate MapReduce jobs'. Together they form a unique fingerprint.

Cite this