DONAR: Decentralized server selection for cloud services

Patrick Wendell, Joe Wenjie Jiang, Michael Joseph Freedman, Jennifer L. Rexford

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

91 Citations (Scopus)

Abstract

Geo-replicated services need an effective way to direct client requests to a particular location, based on performance, load, and cost. This paper presents DONAR, a distributed system that can offload the burden of replica selection, while providing these services with a sufficiently expressive interface for specifying mapping policies. Most existing approaches for replica selection rely on either central coordination (which has reliability, security, and scalability limitations) or distributed heuristics (which lead to suboptimal request distributions, or even instability). In contrast, the distributed mapping nodes in DONAR run a simple, efficient algorithm to coordinate their replica-selection decisions for clients. The protocol solves an optimization problem that jointly considers both client performance and server load, allowing us to show that the distributed algorithm is stable and effective. Experiments with our DONAR prototype - providing replica selection for CoralCDN and the Measurement Lab - demonstrate that our algorithm performs well "in the wild." Our prototype supports DNS- and HTTP-based redirection, IP anycast, and a secure update protocol, and can handle many customer services with diverse policy objectives.

Original languageEnglish (US)
Title of host publicationSIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference
Pages231-242
Number of pages12
DOIs
StatePublished - Nov 15 2010
Event7th International Conference on Autonomic Computing, SIGCOMM 2010 - New Delhi, India
Duration: Aug 30 2010Sep 3 2010

Publication series

NameSIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference

Other

Other7th International Conference on Autonomic Computing, SIGCOMM 2010
CountryIndia
CityNew Delhi
Period8/30/109/3/10

Fingerprint

Replica
Decentralized
Servers
Server
Network protocols
HTTP
Parallel algorithms
Scalability
Prototype
Distributed Algorithms
Costs
Distributed Systems
Efficient Algorithms
Customers
Update
Experiments
Heuristics
Optimization Problem
Vertex of a graph
Demonstrate

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computational Theory and Mathematics

Cite this

Wendell, P., Jiang, J. W., Freedman, M. J., & Rexford, J. L. (2010). DONAR: Decentralized server selection for cloud services. In SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference (pp. 231-242). (SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference). https://doi.org/10.1145/1851182.1851211
Wendell, Patrick ; Jiang, Joe Wenjie ; Freedman, Michael Joseph ; Rexford, Jennifer L. / DONAR : Decentralized server selection for cloud services. SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference. 2010. pp. 231-242 (SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference).
@inproceedings{1e1935099045498294f6b558d665d8aa,
title = "DONAR: Decentralized server selection for cloud services",
abstract = "Geo-replicated services need an effective way to direct client requests to a particular location, based on performance, load, and cost. This paper presents DONAR, a distributed system that can offload the burden of replica selection, while providing these services with a sufficiently expressive interface for specifying mapping policies. Most existing approaches for replica selection rely on either central coordination (which has reliability, security, and scalability limitations) or distributed heuristics (which lead to suboptimal request distributions, or even instability). In contrast, the distributed mapping nodes in DONAR run a simple, efficient algorithm to coordinate their replica-selection decisions for clients. The protocol solves an optimization problem that jointly considers both client performance and server load, allowing us to show that the distributed algorithm is stable and effective. Experiments with our DONAR prototype - providing replica selection for CoralCDN and the Measurement Lab - demonstrate that our algorithm performs well {"}in the wild.{"} Our prototype supports DNS- and HTTP-based redirection, IP anycast, and a secure update protocol, and can handle many customer services with diverse policy objectives.",
author = "Patrick Wendell and Jiang, {Joe Wenjie} and Freedman, {Michael Joseph} and Rexford, {Jennifer L.}",
year = "2010",
month = "11",
day = "15",
doi = "https://doi.org/10.1145/1851182.1851211",
language = "English (US)",
isbn = "9781450302012",
series = "SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference",
pages = "231--242",
booktitle = "SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference",

}

Wendell, P, Jiang, JW, Freedman, MJ & Rexford, JL 2010, DONAR: Decentralized server selection for cloud services. in SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference. SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference, pp. 231-242, 7th International Conference on Autonomic Computing, SIGCOMM 2010, New Delhi, India, 8/30/10. https://doi.org/10.1145/1851182.1851211

DONAR : Decentralized server selection for cloud services. / Wendell, Patrick; Jiang, Joe Wenjie; Freedman, Michael Joseph; Rexford, Jennifer L.

SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference. 2010. p. 231-242 (SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference).

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

TY - GEN

T1 - DONAR

T2 - Decentralized server selection for cloud services

AU - Wendell, Patrick

AU - Jiang, Joe Wenjie

AU - Freedman, Michael Joseph

AU - Rexford, Jennifer L.

PY - 2010/11/15

Y1 - 2010/11/15

N2 - Geo-replicated services need an effective way to direct client requests to a particular location, based on performance, load, and cost. This paper presents DONAR, a distributed system that can offload the burden of replica selection, while providing these services with a sufficiently expressive interface for specifying mapping policies. Most existing approaches for replica selection rely on either central coordination (which has reliability, security, and scalability limitations) or distributed heuristics (which lead to suboptimal request distributions, or even instability). In contrast, the distributed mapping nodes in DONAR run a simple, efficient algorithm to coordinate their replica-selection decisions for clients. The protocol solves an optimization problem that jointly considers both client performance and server load, allowing us to show that the distributed algorithm is stable and effective. Experiments with our DONAR prototype - providing replica selection for CoralCDN and the Measurement Lab - demonstrate that our algorithm performs well "in the wild." Our prototype supports DNS- and HTTP-based redirection, IP anycast, and a secure update protocol, and can handle many customer services with diverse policy objectives.

AB - Geo-replicated services need an effective way to direct client requests to a particular location, based on performance, load, and cost. This paper presents DONAR, a distributed system that can offload the burden of replica selection, while providing these services with a sufficiently expressive interface for specifying mapping policies. Most existing approaches for replica selection rely on either central coordination (which has reliability, security, and scalability limitations) or distributed heuristics (which lead to suboptimal request distributions, or even instability). In contrast, the distributed mapping nodes in DONAR run a simple, efficient algorithm to coordinate their replica-selection decisions for clients. The protocol solves an optimization problem that jointly considers both client performance and server load, allowing us to show that the distributed algorithm is stable and effective. Experiments with our DONAR prototype - providing replica selection for CoralCDN and the Measurement Lab - demonstrate that our algorithm performs well "in the wild." Our prototype supports DNS- and HTTP-based redirection, IP anycast, and a secure update protocol, and can handle many customer services with diverse policy objectives.

UR - http://www.scopus.com/inward/record.url?scp=78149301795&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78149301795&partnerID=8YFLogxK

U2 - https://doi.org/10.1145/1851182.1851211

DO - https://doi.org/10.1145/1851182.1851211

M3 - Conference contribution

SN - 9781450302012

T3 - SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference

SP - 231

EP - 242

BT - SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference

ER -

Wendell P, Jiang JW, Freedman MJ, Rexford JL. DONAR: Decentralized server selection for cloud services. In SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference. 2010. p. 231-242. (SIGCOMM'10 - Proceedings of the SIGCOMM 2010 Conference). https://doi.org/10.1145/1851182.1851211