Graphless: Toward serverless graph processing

Lucian Toader, Alexandru Uta, Ahmed Musaafir, Alexandru Iosup

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

Our society is increasingly solving complex problems through the use of graph processing. Existing graph processing systems focus on performance, which allows addressing ever-larger and more complex problems. They also require uncommon expertise to properly deploy and utilize. To make graph processing generally accessible-to small and medium enterprises and institutions, to common research groups, to individuals-, in this work we design and implement the Graphless graph-processing system. Graphless is based on the serverless paradigm, which proposes to simplify computing by letting developers only focus on small, stateless functions, which are deployed and managed automatically. We address with Graphless the key challenge of combining the stateless functions assumed by serverless computing with the (opposite) data-intensive nature of graph processing. Graphless tackles this challenge through an architectural approach that allows it to deploy with push or with pull operation, and a collection of backend services, such as an orchestrator and a memory-as-a-service component. We implement Graphless and conduct with it real-world experiments using Amazon Lambda for cloud-based serverless resources. Using the LDBC Graphalytics benchmark, we analyze Graphless, and compare its performance and operational cost with the graph-processing systems Apache Giraph (big data domain) and GraphMat (HPC). Overall, we show evidence Graphless provides performance and cost-efficiency similar to Giraph, for algorithms that can benefit from fine-grained elasticity, and lower than GraphMat, but is architecturally easier to deploy, and provides both push and pull operation.

Original languageEnglish
Title of host publication2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings
EditorsAlexandru Iosup, Radu Prodan, Alexandru Uta, Florin Pop
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages66-73
Number of pages8
ISBN (Electronic)9781728138008
DOIs
Publication statusPublished - 3 Jun 2019
Event18th International Symposium on Parallel and Distributed Computing, ISPDC 2019 - Amsterdam, Netherlands
Duration: 5 Jun 20197 Jun 2019

Conference

Conference18th International Symposium on Parallel and Distributed Computing, ISPDC 2019
CountryNetherlands
CityAmsterdam
Period5/06/197/06/19

Fingerprint

Processing
Graph
Costs
Elasticity
Data storage equipment
Industry
Experiments
Pull

Keywords

  • Elasticity
  • Function as a Service
  • Graph processing
  • Serverless

Cite this

Toader, L., Uta, A., Musaafir, A., & Iosup, A. (2019). Graphless: Toward serverless graph processing. In A. Iosup, R. Prodan, A. Uta, & F. Pop (Eds.), 2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings (pp. 66-73). [8790945] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISPDC.2019.00012
Toader, Lucian ; Uta, Alexandru ; Musaafir, Ahmed ; Iosup, Alexandru. / Graphless : Toward serverless graph processing. 2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings. editor / Alexandru Iosup ; Radu Prodan ; Alexandru Uta ; Florin Pop. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 66-73
@inproceedings{c781d284f9ec4f77ae5056a94ee2577c,
title = "Graphless: Toward serverless graph processing",
abstract = "Our society is increasingly solving complex problems through the use of graph processing. Existing graph processing systems focus on performance, which allows addressing ever-larger and more complex problems. They also require uncommon expertise to properly deploy and utilize. To make graph processing generally accessible-to small and medium enterprises and institutions, to common research groups, to individuals-, in this work we design and implement the Graphless graph-processing system. Graphless is based on the serverless paradigm, which proposes to simplify computing by letting developers only focus on small, stateless functions, which are deployed and managed automatically. We address with Graphless the key challenge of combining the stateless functions assumed by serverless computing with the (opposite) data-intensive nature of graph processing. Graphless tackles this challenge through an architectural approach that allows it to deploy with push or with pull operation, and a collection of backend services, such as an orchestrator and a memory-as-a-service component. We implement Graphless and conduct with it real-world experiments using Amazon Lambda for cloud-based serverless resources. Using the LDBC Graphalytics benchmark, we analyze Graphless, and compare its performance and operational cost with the graph-processing systems Apache Giraph (big data domain) and GraphMat (HPC). Overall, we show evidence Graphless provides performance and cost-efficiency similar to Giraph, for algorithms that can benefit from fine-grained elasticity, and lower than GraphMat, but is architecturally easier to deploy, and provides both push and pull operation.",
keywords = "Elasticity, Function as a Service, Graph processing, Serverless",
author = "Lucian Toader and Alexandru Uta and Ahmed Musaafir and Alexandru Iosup",
year = "2019",
month = "6",
day = "3",
doi = "10.1109/ISPDC.2019.00012",
language = "English",
pages = "66--73",
editor = "Alexandru Iosup and Radu Prodan and Alexandru Uta and Florin Pop",
booktitle = "2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

Toader, L, Uta, A, Musaafir, A & Iosup, A 2019, Graphless: Toward serverless graph processing. in A Iosup, R Prodan, A Uta & F Pop (eds), 2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings., 8790945, Institute of Electrical and Electronics Engineers Inc., pp. 66-73, 18th International Symposium on Parallel and Distributed Computing, ISPDC 2019, Amsterdam, Netherlands, 5/06/19. https://doi.org/10.1109/ISPDC.2019.00012

Graphless : Toward serverless graph processing. / Toader, Lucian; Uta, Alexandru; Musaafir, Ahmed; Iosup, Alexandru.

2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings. ed. / Alexandru Iosup; Radu Prodan; Alexandru Uta; Florin Pop. Institute of Electrical and Electronics Engineers Inc., 2019. p. 66-73 8790945.

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Graphless

T2 - Toward serverless graph processing

AU - Toader, Lucian

AU - Uta, Alexandru

AU - Musaafir, Ahmed

AU - Iosup, Alexandru

PY - 2019/6/3

Y1 - 2019/6/3

N2 - Our society is increasingly solving complex problems through the use of graph processing. Existing graph processing systems focus on performance, which allows addressing ever-larger and more complex problems. They also require uncommon expertise to properly deploy and utilize. To make graph processing generally accessible-to small and medium enterprises and institutions, to common research groups, to individuals-, in this work we design and implement the Graphless graph-processing system. Graphless is based on the serverless paradigm, which proposes to simplify computing by letting developers only focus on small, stateless functions, which are deployed and managed automatically. We address with Graphless the key challenge of combining the stateless functions assumed by serverless computing with the (opposite) data-intensive nature of graph processing. Graphless tackles this challenge through an architectural approach that allows it to deploy with push or with pull operation, and a collection of backend services, such as an orchestrator and a memory-as-a-service component. We implement Graphless and conduct with it real-world experiments using Amazon Lambda for cloud-based serverless resources. Using the LDBC Graphalytics benchmark, we analyze Graphless, and compare its performance and operational cost with the graph-processing systems Apache Giraph (big data domain) and GraphMat (HPC). Overall, we show evidence Graphless provides performance and cost-efficiency similar to Giraph, for algorithms that can benefit from fine-grained elasticity, and lower than GraphMat, but is architecturally easier to deploy, and provides both push and pull operation.

AB - Our society is increasingly solving complex problems through the use of graph processing. Existing graph processing systems focus on performance, which allows addressing ever-larger and more complex problems. They also require uncommon expertise to properly deploy and utilize. To make graph processing generally accessible-to small and medium enterprises and institutions, to common research groups, to individuals-, in this work we design and implement the Graphless graph-processing system. Graphless is based on the serverless paradigm, which proposes to simplify computing by letting developers only focus on small, stateless functions, which are deployed and managed automatically. We address with Graphless the key challenge of combining the stateless functions assumed by serverless computing with the (opposite) data-intensive nature of graph processing. Graphless tackles this challenge through an architectural approach that allows it to deploy with push or with pull operation, and a collection of backend services, such as an orchestrator and a memory-as-a-service component. We implement Graphless and conduct with it real-world experiments using Amazon Lambda for cloud-based serverless resources. Using the LDBC Graphalytics benchmark, we analyze Graphless, and compare its performance and operational cost with the graph-processing systems Apache Giraph (big data domain) and GraphMat (HPC). Overall, we show evidence Graphless provides performance and cost-efficiency similar to Giraph, for algorithms that can benefit from fine-grained elasticity, and lower than GraphMat, but is architecturally easier to deploy, and provides both push and pull operation.

KW - Elasticity

KW - Function as a Service

KW - Graph processing

KW - Serverless

UR - http://www.scopus.com/inward/record.url?scp=85071513030&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071513030&partnerID=8YFLogxK

U2 - 10.1109/ISPDC.2019.00012

DO - 10.1109/ISPDC.2019.00012

M3 - Conference contribution

SP - 66

EP - 73

BT - 2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings

A2 - Iosup, Alexandru

A2 - Prodan, Radu

A2 - Uta, Alexandru

A2 - Pop, Florin

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Toader L, Uta A, Musaafir A, Iosup A. Graphless: Toward serverless graph processing. In Iosup A, Prodan R, Uta A, Pop F, editors, 2019 18th International Symposium on Parallel and Distributed Computing (ISPDC) - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2019. p. 66-73. 8790945 https://doi.org/10.1109/ISPDC.2019.00012