Fully decentralized emulation of best-effort and processor sharing queues

Rade Stanojevic, Robert Shorten

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

20 Citations (Scopus)

Abstract

Control of large distributed cloud-based services is a challenging problem. The Distributed Rate Limiting (DRL) paradigm was recently proposed as a mechanism for tackling this problem. The heuristic nature of existing DRL solutions makes their behavior unpredictable and analytically untractable. In this paper we treat the DRL problem in a mathematical framework and propose two novel DRL algorithms that exhibit good and predictable performance. The first algorithm Cloud Control with Constant Probabilities (C3P) solves the DRL problem in best effort environments, emulating the behavior of a single best-effort queue in a fully distributed manner. The second problem we approach is the DRL in processor sharing environments. Our algorithm, Distributed Deficit Round Robin (D2R2), parameterized by parameter α, converges to a state that is, at most, O(1/α) away from the exact emulation of centralized processor sharing queue. The convergence and stability properties are fully analyzed for both C3P and D2R2. Analytical results are validated empirically through a number of representative packet level simulations. The closed-form nature of our results allows simple design rules which, together with extremely low communication overhead, makes the presented algorithms practical and easy to deploy.

Original languageEnglish
Title of host publicationSIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems
Pages383-394
Number of pages12
Volume36
Edition1 SPECIAL ISSUE
DOIs
Publication statusPublished - 2008
Externally publishedYes
Event2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, SIGMETRICS'08 - Annapolis, MD, United States
Duration: 2 Jun 20086 Jun 2008

Other

Other2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, SIGMETRICS'08
CountryUnited States
CityAnnapolis, MD
Period2/6/086/6/08

Fingerprint

Parallel algorithms
Communication

Keywords

  • CDK
  • Cloud control
  • Consensus agreement
  • Rate limiting
  • Stability and convergence

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Software

Cite this

Stanojevic, R., & Shorten, R. (2008). Fully decentralized emulation of best-effort and processor sharing queues. In SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems (1 SPECIAL ISSUE ed., Vol. 36, pp. 383-394) https://doi.org/10.1145/1375457.1375501

Fully decentralized emulation of best-effort and processor sharing queues. / Stanojevic, Rade; Shorten, Robert.

SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems. Vol. 36 1 SPECIAL ISSUE. ed. 2008. p. 383-394.

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

Stanojevic, R & Shorten, R 2008, Fully decentralized emulation of best-effort and processor sharing queues. in SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems. 1 SPECIAL ISSUE edn, vol. 36, pp. 383-394, 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, SIGMETRICS'08, Annapolis, MD, United States, 2/6/08. https://doi.org/10.1145/1375457.1375501
Stanojevic R, Shorten R. Fully decentralized emulation of best-effort and processor sharing queues. In SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems. 1 SPECIAL ISSUE ed. Vol. 36. 2008. p. 383-394 https://doi.org/10.1145/1375457.1375501
Stanojevic, Rade ; Shorten, Robert. / Fully decentralized emulation of best-effort and processor sharing queues. SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems. Vol. 36 1 SPECIAL ISSUE. ed. 2008. pp. 383-394
@inproceedings{456cf955240c4cf7892239055b22df4c,
title = "Fully decentralized emulation of best-effort and processor sharing queues",
abstract = "Control of large distributed cloud-based services is a challenging problem. The Distributed Rate Limiting (DRL) paradigm was recently proposed as a mechanism for tackling this problem. The heuristic nature of existing DRL solutions makes their behavior unpredictable and analytically untractable. In this paper we treat the DRL problem in a mathematical framework and propose two novel DRL algorithms that exhibit good and predictable performance. The first algorithm Cloud Control with Constant Probabilities (C3P) solves the DRL problem in best effort environments, emulating the behavior of a single best-effort queue in a fully distributed manner. The second problem we approach is the DRL in processor sharing environments. Our algorithm, Distributed Deficit Round Robin (D2R2), parameterized by parameter α, converges to a state that is, at most, O(1/α) away from the exact emulation of centralized processor sharing queue. The convergence and stability properties are fully analyzed for both C3P and D2R2. Analytical results are validated empirically through a number of representative packet level simulations. The closed-form nature of our results allows simple design rules which, together with extremely low communication overhead, makes the presented algorithms practical and easy to deploy.",
keywords = "CDK, Cloud control, Consensus agreement, Rate limiting, Stability and convergence",
author = "Rade Stanojevic and Robert Shorten",
year = "2008",
doi = "10.1145/1375457.1375501",
language = "English",
isbn = "9781605580050",
volume = "36",
pages = "383--394",
booktitle = "SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems",
edition = "1 SPECIAL ISSUE",

}

TY - GEN

T1 - Fully decentralized emulation of best-effort and processor sharing queues

AU - Stanojevic, Rade

AU - Shorten, Robert

PY - 2008

Y1 - 2008

N2 - Control of large distributed cloud-based services is a challenging problem. The Distributed Rate Limiting (DRL) paradigm was recently proposed as a mechanism for tackling this problem. The heuristic nature of existing DRL solutions makes their behavior unpredictable and analytically untractable. In this paper we treat the DRL problem in a mathematical framework and propose two novel DRL algorithms that exhibit good and predictable performance. The first algorithm Cloud Control with Constant Probabilities (C3P) solves the DRL problem in best effort environments, emulating the behavior of a single best-effort queue in a fully distributed manner. The second problem we approach is the DRL in processor sharing environments. Our algorithm, Distributed Deficit Round Robin (D2R2), parameterized by parameter α, converges to a state that is, at most, O(1/α) away from the exact emulation of centralized processor sharing queue. The convergence and stability properties are fully analyzed for both C3P and D2R2. Analytical results are validated empirically through a number of representative packet level simulations. The closed-form nature of our results allows simple design rules which, together with extremely low communication overhead, makes the presented algorithms practical and easy to deploy.

AB - Control of large distributed cloud-based services is a challenging problem. The Distributed Rate Limiting (DRL) paradigm was recently proposed as a mechanism for tackling this problem. The heuristic nature of existing DRL solutions makes their behavior unpredictable and analytically untractable. In this paper we treat the DRL problem in a mathematical framework and propose two novel DRL algorithms that exhibit good and predictable performance. The first algorithm Cloud Control with Constant Probabilities (C3P) solves the DRL problem in best effort environments, emulating the behavior of a single best-effort queue in a fully distributed manner. The second problem we approach is the DRL in processor sharing environments. Our algorithm, Distributed Deficit Round Robin (D2R2), parameterized by parameter α, converges to a state that is, at most, O(1/α) away from the exact emulation of centralized processor sharing queue. The convergence and stability properties are fully analyzed for both C3P and D2R2. Analytical results are validated empirically through a number of representative packet level simulations. The closed-form nature of our results allows simple design rules which, together with extremely low communication overhead, makes the presented algorithms practical and easy to deploy.

KW - CDK

KW - Cloud control

KW - Consensus agreement

KW - Rate limiting

KW - Stability and convergence

UR - http://www.scopus.com/inward/record.url?scp=57449089029&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=57449089029&partnerID=8YFLogxK

U2 - 10.1145/1375457.1375501

DO - 10.1145/1375457.1375501

M3 - Conference contribution

AN - SCOPUS:57449089029

SN - 9781605580050

VL - 36

SP - 383

EP - 394

BT - SIGMETRICS'08: Proceedings of the 2008 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems

ER -