A lightweight communication runtime for distributed graph analytics

Hoang Vu Dang; Roshan Dathathri; Gurbinder Gill; Alex Brooks; Nikoli Dryden; Andrew Lenharth; Loc Hoang; Keshav Pingali; Marc Snir

doi:10.1109/IPDPS.2018.00107

A lightweight communication runtime for distributed graph analytics

Hoang Vu Dang, Roshan Dathathri, Gurbinder Gill, Alex Brooks, Nikoli Dryden, Andrew Lenharth, Loc Hoang, Keshav Pingali, Marc Snir

Siebel School of Computing and Data Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Distributed-memory multi-core clusters enable in-memory processing of very large graphs with billions of nodes and edges. Recent distributed graph analytics systems have been built on top of MPI. However, communication in graph applications is very irregular, and each host exchanges different amounts of non-contiguous data with other hosts. MPI does not support such a communication pattern well, and it has limited ability to integrate communication with serialization, deserialization, and graph computation tasks. In this paper, we describe a lightweight communication runtime called LCI that supports a large number of threads on each host and avoids the semantic mismatches between the requirements of graph computations and the communication library in MPI. The implementation of LCI is informed by lessons learnt from two baseline MPI-based implementations. We have successfully integrated LCI with two state-of-The-Art graph analytics systems-Gemini and Abelian. LCI improves the latency up to 3.5x for microbenchmarks compared to MPI solutions and improves the end-To-end performance of distributed graph algorithms by up to 2x.

Original language	English (US)
Title of host publication	Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	980-989
Number of pages	10
ISBN (Print)	9781538643686
DOIs	https://doi.org/10.1109/IPDPS.2018.00107
State	Published - Aug 3 2018
Event	32nd IEEE International Parallel and Distributed Processing Symposium, IPDPS 2018 - Vancouver, Canada Duration: May 21 2018 → May 25 2018

Publication series

Name	Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018

Other

Other	32nd IEEE International Parallel and Distributed Processing Symposium, IPDPS 2018
Country/Territory	Canada
City	Vancouver
Period	5/21/18 → 5/25/18

Keywords

Clusters
Communication systems
Graph analytics

ASJC Scopus subject areas

Artificial Intelligence
Computer Networks and Communications
Hardware and Architecture
Information Systems and Management

Online availability

10.1109/IPDPS.2018.00107

Library availability

Discover UIUC Full Text

Cite this

Dang, H. V., Dathathri, R., Gill, G., Brooks, A., Dryden, N., Lenharth, A., Hoang, L., Pingali, K., & Snir, M. (2018). A lightweight communication runtime for distributed graph analytics. In Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018 (pp. 980-989). Article 8425251 (Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPDPS.2018.00107

A lightweight communication runtime for distributed graph analytics. / Dang, Hoang Vu; Dathathri, Roshan; Gill, Gurbinder et al.
Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 980-989 8425251 (Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Dang, HV, Dathathri, R, Gill, G, Brooks, A, Dryden, N, Lenharth, A, Hoang, L, Pingali, K & Snir, M 2018, A lightweight communication runtime for distributed graph analytics. in Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018., 8425251, Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018, Institute of Electrical and Electronics Engineers Inc., pp. 980-989, 32nd IEEE International Parallel and Distributed Processing Symposium, IPDPS 2018, Vancouver, Canada, 5/21/18. https://doi.org/10.1109/IPDPS.2018.00107

Dang HV, Dathathri R, Gill G, Brooks A, Dryden N, Lenharth A et al. A lightweight communication runtime for distributed graph analytics. In Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 980-989. 8425251. (Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018). doi: 10.1109/IPDPS.2018.00107

Dang, Hoang Vu ; Dathathri, Roshan ; Gill, Gurbinder et al. / A lightweight communication runtime for distributed graph analytics. Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 980-989 (Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018).

@inproceedings{6356a57bd94046c8adb8c5c22adbd690,

title = "A lightweight communication runtime for distributed graph analytics",

abstract = "Distributed-memory multi-core clusters enable in-memory processing of very large graphs with billions of nodes and edges. Recent distributed graph analytics systems have been built on top of MPI. However, communication in graph applications is very irregular, and each host exchanges different amounts of non-contiguous data with other hosts. MPI does not support such a communication pattern well, and it has limited ability to integrate communication with serialization, deserialization, and graph computation tasks. In this paper, we describe a lightweight communication runtime called LCI that supports a large number of threads on each host and avoids the semantic mismatches between the requirements of graph computations and the communication library in MPI. The implementation of LCI is informed by lessons learnt from two baseline MPI-based implementations. We have successfully integrated LCI with two state-of-The-Art graph analytics systems-Gemini and Abelian. LCI improves the latency up to 3.5x for microbenchmarks compared to MPI solutions and improves the end-To-end performance of distributed graph algorithms by up to 2x.",

keywords = "Clusters, Communication systems, Graph analytics",

author = "Dang, {Hoang Vu} and Roshan Dathathri and Gurbinder Gill and Alex Brooks and Nikoli Dryden and Andrew Lenharth and Loc Hoang and Keshav Pingali and Marc Snir",

note = "Publisher Copyright: {\textcopyright} 2018 IEEE.; 32nd IEEE International Parallel and Distributed Processing Symposium, IPDPS 2018 ; Conference date: 21-05-2018 Through 25-05-2018",

year = "2018",

month = aug,

day = "3",

doi = "10.1109/IPDPS.2018.00107",

language = "English (US)",

isbn = "9781538643686",

series = "Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "980--989",

booktitle = "Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018",

address = "United States",

}

TY - GEN

T1 - A lightweight communication runtime for distributed graph analytics

AU - Dang, Hoang Vu

AU - Dathathri, Roshan

AU - Gill, Gurbinder

AU - Brooks, Alex

AU - Dryden, Nikoli

AU - Lenharth, Andrew

AU - Hoang, Loc

AU - Pingali, Keshav

AU - Snir, Marc

PY - 2018/8/3

Y1 - 2018/8/3

N2 - Distributed-memory multi-core clusters enable in-memory processing of very large graphs with billions of nodes and edges. Recent distributed graph analytics systems have been built on top of MPI. However, communication in graph applications is very irregular, and each host exchanges different amounts of non-contiguous data with other hosts. MPI does not support such a communication pattern well, and it has limited ability to integrate communication with serialization, deserialization, and graph computation tasks. In this paper, we describe a lightweight communication runtime called LCI that supports a large number of threads on each host and avoids the semantic mismatches between the requirements of graph computations and the communication library in MPI. The implementation of LCI is informed by lessons learnt from two baseline MPI-based implementations. We have successfully integrated LCI with two state-of-The-Art graph analytics systems-Gemini and Abelian. LCI improves the latency up to 3.5x for microbenchmarks compared to MPI solutions and improves the end-To-end performance of distributed graph algorithms by up to 2x.

AB - Distributed-memory multi-core clusters enable in-memory processing of very large graphs with billions of nodes and edges. Recent distributed graph analytics systems have been built on top of MPI. However, communication in graph applications is very irregular, and each host exchanges different amounts of non-contiguous data with other hosts. MPI does not support such a communication pattern well, and it has limited ability to integrate communication with serialization, deserialization, and graph computation tasks. In this paper, we describe a lightweight communication runtime called LCI that supports a large number of threads on each host and avoids the semantic mismatches between the requirements of graph computations and the communication library in MPI. The implementation of LCI is informed by lessons learnt from two baseline MPI-based implementations. We have successfully integrated LCI with two state-of-The-Art graph analytics systems-Gemini and Abelian. LCI improves the latency up to 3.5x for microbenchmarks compared to MPI solutions and improves the end-To-end performance of distributed graph algorithms by up to 2x.

KW - Clusters

KW - Communication systems

KW - Graph analytics

UR - http://www.scopus.com/inward/record.url?scp=85049588856&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049588856&partnerID=8YFLogxK

U2 - 10.1109/IPDPS.2018.00107

DO - 10.1109/IPDPS.2018.00107

M3 - Conference contribution

AN - SCOPUS:85049588856

SN - 9781538643686

T3 - Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018

SP - 980

EP - 989

BT - Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 32nd IEEE International Parallel and Distributed Processing Symposium, IPDPS 2018

Y2 - 21 May 2018 through 25 May 2018

ER -

A lightweight communication runtime for distributed graph analytics

Abstract

Publication series

Other

Keywords

ASJC Scopus subject areas

Online availability

Library availability

Related links

Fingerprint

Cite this