New techniques to curtail the tail latency in stream processing systems

Guangxiang Du, Indranil Gupta

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

Abstract

This paper presents a series of novel techniques for reducing the tail latency in stream processing systems like Apache Storm. Concretely, we present three mechanisms: (1) adaptive timeout coupled with selective replay to catch straggler tuples; (2) shared queues among different tasks of the same operator to reduce overall queueing delay; (3) latency feedback-based load balancing, intended to mitigate het-erogenous scenarios. We have implemented these techniques in Apache Storm, and present experimental results using sets of micro-benchmarks as well as two topologies from Yahoo! Inc. Our results show improvement in tail latency up to 72.9%.

Original languageEnglish (US)
Title of host publicationProceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016
PublisherAssociation for Computing Machinery
ISBN (Print)9781450342209
DOIs
StatePublished - Jul 25 2016
Event4th Annual ACM PODC Workshop on Distributed Cloud Computing, DCC 2016 - Chicago, United States
Duration: Jul 25 2016Jul 28 2016

Publication series

NameProceedings of the Annual ACM Symposium on Principles of Distributed Computing

Other

Other4th Annual ACM PODC Workshop on Distributed Cloud Computing, DCC 2016
CountryUnited States
CityChicago
Period7/25/167/28/16

Fingerprint

Processing
Resource allocation
Topology
Feedback

Keywords

  • Apache Storm
  • Stream Processing Systems
  • Tail Latency

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Computer Networks and Communications

Cite this

Du, G., & Gupta, I. (2016). New techniques to curtail the tail latency in stream processing systems In Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016 [a7] (Proceedings of the Annual ACM Symposium on Principles of Distributed Computing). Association for Computing Machinery. https://doi.org/10.1145/2955193.2955206

New techniques to curtail the tail latency in stream processing systems . / Du, Guangxiang; Gupta, Indranil.

Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016. Association for Computing Machinery, 2016. a7 (Proceedings of the Annual ACM Symposium on Principles of Distributed Computing).

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

Du, G & Gupta, I 2016, New techniques to curtail the tail latency in stream processing systems in Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016., a7, Proceedings of the Annual ACM Symposium on Principles of Distributed Computing, Association for Computing Machinery, 4th Annual ACM PODC Workshop on Distributed Cloud Computing, DCC 2016, Chicago, United States, 7/25/16. https://doi.org/10.1145/2955193.2955206
Du G, Gupta I. New techniques to curtail the tail latency in stream processing systems In Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016. Association for Computing Machinery. 2016. a7. (Proceedings of the Annual ACM Symposium on Principles of Distributed Computing). https://doi.org/10.1145/2955193.2955206
Du, Guangxiang ; Gupta, Indranil. / New techniques to curtail the tail latency in stream processing systems Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016. Association for Computing Machinery, 2016. (Proceedings of the Annual ACM Symposium on Principles of Distributed Computing).
@inproceedings{4431fd4ec55349bebaea64366fa9d48d,
title = "New techniques to curtail the tail latency in stream processing systems ∗",
abstract = "This paper presents a series of novel techniques for reducing the tail latency in stream processing systems like Apache Storm. Concretely, we present three mechanisms: (1) adaptive timeout coupled with selective replay to catch straggler tuples; (2) shared queues among different tasks of the same operator to reduce overall queueing delay; (3) latency feedback-based load balancing, intended to mitigate het-erogenous scenarios. We have implemented these techniques in Apache Storm, and present experimental results using sets of micro-benchmarks as well as two topologies from Yahoo! Inc. Our results show improvement in tail latency up to 72.9{\%}.",
keywords = "Apache Storm, Stream Processing Systems, Tail Latency",
author = "Guangxiang Du and Indranil Gupta",
year = "2016",
month = "7",
day = "25",
doi = "10.1145/2955193.2955206",
language = "English (US)",
isbn = "9781450342209",
series = "Proceedings of the Annual ACM Symposium on Principles of Distributed Computing",
publisher = "Association for Computing Machinery",
booktitle = "Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016",

}

TY - GEN

T1 - New techniques to curtail the tail latency in stream processing systems ∗

AU - Du, Guangxiang

AU - Gupta, Indranil

PY - 2016/7/25

Y1 - 2016/7/25

N2 - This paper presents a series of novel techniques for reducing the tail latency in stream processing systems like Apache Storm. Concretely, we present three mechanisms: (1) adaptive timeout coupled with selective replay to catch straggler tuples; (2) shared queues among different tasks of the same operator to reduce overall queueing delay; (3) latency feedback-based load balancing, intended to mitigate het-erogenous scenarios. We have implemented these techniques in Apache Storm, and present experimental results using sets of micro-benchmarks as well as two topologies from Yahoo! Inc. Our results show improvement in tail latency up to 72.9%.

AB - This paper presents a series of novel techniques for reducing the tail latency in stream processing systems like Apache Storm. Concretely, we present three mechanisms: (1) adaptive timeout coupled with selective replay to catch straggler tuples; (2) shared queues among different tasks of the same operator to reduce overall queueing delay; (3) latency feedback-based load balancing, intended to mitigate het-erogenous scenarios. We have implemented these techniques in Apache Storm, and present experimental results using sets of micro-benchmarks as well as two topologies from Yahoo! Inc. Our results show improvement in tail latency up to 72.9%.

KW - Apache Storm

KW - Stream Processing Systems

KW - Tail Latency

UR - http://www.scopus.com/inward/record.url?scp=85046652281&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85046652281&partnerID=8YFLogxK

U2 - 10.1145/2955193.2955206

DO - 10.1145/2955193.2955206

M3 - Conference contribution

AN - SCOPUS:85046652281

SN - 9781450342209

T3 - Proceedings of the Annual ACM Symposium on Principles of Distributed Computing

BT - Proceedings of the 4th Workshop on Distributed Cloud Computing, DCC 2016

PB - Association for Computing Machinery

ER -