In-Network Computation is a Dumb Idea Whose Time Has Come

Handle URI:
http://hdl.handle.net/10754/627118
Title:
In-Network Computation is a Dumb Idea Whose Time Has Come
Authors:
Sapio, Amedeo; Abdelaziz, Ibrahim ( 0000-0003-1449-5115 ) ; Aldilaijan, Abdulla; Canini, Marco ( 0000-0002-5051-4283 ) ; Kalnis, Panos ( 0000-0002-5060-1360 )
Abstract:
Programmable data plane hardware creates new opportunities for infusing intelligence into the network. This raises a fundamental question: what kinds of computation should be delegated to the network? In this paper, we discuss the opportunities and challenges for co-designing data center distributed systems with their network layer. We believe that the time has finally come for offloading part of their computation to execute in-network. However, in-network computation tasks must be judiciously crafted to match the limitations of the network machine architecture of programmable devices. With the help of our experiments on machine learning and graph analytics workloads, we identify that aggregation functions raise opportunities to exploit the limited computation power of networking hardware to lessen network congestion and improve the overall application performance. Moreover, as a proof-of-concept, we propose DAIET, a system that performs in-network data aggregation. Experimental results with an initial prototype show a large data reduction ratio (86.9%-89.3%) and a similar decrease in the workers' computation time.
KAUST Department:
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division; Computer Science Program
Citation:
Sapio A, Abdelaziz I, Aldilaijan A, Canini M, Kalnis P (2017) In-Network Computation is a Dumb Idea Whose Time Has Come. Proceedings of the 16th ACM Workshop on Hot Topics in Networks - HotNets-XVI. Available: http://dx.doi.org/10.1145/3152434.3152461.
Publisher:
ACM Press
Journal:
Proceedings of the 16th ACM Workshop on Hot Topics in Networks - HotNets-XVI
Conference/Event name:
16th ACM Workshop on Hot Topics in Networks, HotNets 2017
Issue Date:
27-Nov-2017
DOI:
10.1145/3152434.3152461
Type:
Conference Paper
Sponsors:
We thank the anonymous reviewers for their feedback. We are grateful to Colin Dixon, Changhoon Kim, Jeongkeun Lee, Jeff Mogul, KyoungSoo Park and Amin Vahdat for their valuable comments and suggestions. We further thank Jeff for inspiring the title of this paper.
Additional Links:
https://dl.acm.org/citation.cfm?doid=3152434.3152461
Appears in Collections:
Conference Papers; Computer Science Program; Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division

Full metadata record

DC FieldValue Language
dc.contributor.authorSapio, Amedeoen
dc.contributor.authorAbdelaziz, Ibrahimen
dc.contributor.authorAldilaijan, Abdullaen
dc.contributor.authorCanini, Marcoen
dc.contributor.authorKalnis, Panosen
dc.date.accessioned2018-02-13T13:43:18Z-
dc.date.available2018-02-13T13:43:18Z-
dc.date.issued2017-11-27en
dc.identifier.citationSapio A, Abdelaziz I, Aldilaijan A, Canini M, Kalnis P (2017) In-Network Computation is a Dumb Idea Whose Time Has Come. Proceedings of the 16th ACM Workshop on Hot Topics in Networks - HotNets-XVI. Available: http://dx.doi.org/10.1145/3152434.3152461.en
dc.identifier.doi10.1145/3152434.3152461en
dc.identifier.urihttp://hdl.handle.net/10754/627118-
dc.description.abstractProgrammable data plane hardware creates new opportunities for infusing intelligence into the network. This raises a fundamental question: what kinds of computation should be delegated to the network? In this paper, we discuss the opportunities and challenges for co-designing data center distributed systems with their network layer. We believe that the time has finally come for offloading part of their computation to execute in-network. However, in-network computation tasks must be judiciously crafted to match the limitations of the network machine architecture of programmable devices. With the help of our experiments on machine learning and graph analytics workloads, we identify that aggregation functions raise opportunities to exploit the limited computation power of networking hardware to lessen network congestion and improve the overall application performance. Moreover, as a proof-of-concept, we propose DAIET, a system that performs in-network data aggregation. Experimental results with an initial prototype show a large data reduction ratio (86.9%-89.3%) and a similar decrease in the workers' computation time.en
dc.description.sponsorshipWe thank the anonymous reviewers for their feedback. We are grateful to Colin Dixon, Changhoon Kim, Jeongkeun Lee, Jeff Mogul, KyoungSoo Park and Amin Vahdat for their valuable comments and suggestions. We further thank Jeff for inspiring the title of this paper.en
dc.publisherACM Pressen
dc.relation.urlhttps://dl.acm.org/citation.cfm?doid=3152434.3152461en
dc.rightsPermission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org.en
dc.titleIn-Network Computation is a Dumb Idea Whose Time Has Comeen
dc.typeConference Paperen
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Divisionen
dc.contributor.departmentComputer Science Programen
dc.identifier.journalProceedings of the 16th ACM Workshop on Hot Topics in Networks - HotNets-XVIen
dc.conference.date2017-11-30 to 2017-12-01en
dc.conference.name16th ACM Workshop on Hot Topics in Networks, HotNets 2017en
dc.conference.locationPalo Alto, CA, USAen
dc.eprint.versionPublisher's Version/PDFen
kaust.authorSapio, Amedeoen
kaust.authorAbdelaziz, Ibrahimen
kaust.authorAldilaijan, Abdullaen
kaust.authorCanini, Marcoen
kaust.authorKalnis, Panosen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.