ScootR: Scaling R Dataframes on Dataflow Systems

Research output: Chapter in Book/Report/Conference proceedingConference contribution

  • External authors:
  • Andreas Kunft
  • Lukas Stadler
  • Daniele Bonetta
  • Cosmin Basca
  • Jens Meiners
  • Sebastian Bress
  • Tilmann Rabl
  • Volker Markl

Abstract

To cope with today's large scale of data, parallel dataflow engines such as Hadoop, and more recently Spark and Flink, have been proposed. They offer scalability and performance, but require data scientists to develop analysis pipelines in unfamiliar programming languages and abstractions. To overcome this hurdle, dataflow engines have introduced some forms of multi-language integrations, e.g., for Python and R. However, this results in data exchange between the dataflow engine and the integrated language runtime, which requires inter-process communication and causes high runtime overheads. In this paper, we present ScootR, a novel approach to execute R in dataflow systems. ScootR tightly integrates the dataflow and R language runtime by using the Truffle framework and the Graal compiler. As a result, ScootR executes R scripts directly in the Flink data processing engine, without serialization and inter-process communication. Our experimental study reveals that ScootR outperforms state-of-the-art systems by up to an order of magnitude.

Bibliographical metadata

Original languageEnglish
Title of host publicationScootR: Scaling R Dataframes on Dataflow Systems
ISBN (Electronic)978-1-4503-6011-1
DOIs
Publication statusPublished - 11 Nov 2018
EventACM Symposium on Cloud Computing 2018 - Cape Rey Beach Resort, Carlsbad, California, Carlsbad, United States
Event duration: 11 Oct 201813 Oct 2018
https://acmsocc.github.io/2018/

Conference

ConferenceACM Symposium on Cloud Computing 2018
CountryUnited States
CityCarlsbad
Period11/10/1813/10/18
Internet address