Abstract
This document examines five performance questions which are repeatedly asked by practitioners in industry: (i) My system utilization is very low, so why are job delays so high? (ii) What should I do to lower job delays? (iii) How can I favor short jobs if I don't know which jobs are short? (iv) If some jobs are more important than others, how do I negotiate importance versus size? (v) How do answers change when dealing with a closed-loop system, rather than an open system? All these questions have simple answers through queueing theory. This short paper elaborates on the questions and their answers. To keep things readable, our tone is purposely informal throughout. For more formal statements of these questions and answers, please see [14].
- Samuli Aalto and Urtzi Ayesta. Optimal scheduling of jobs with a DHR tail in the M/G/1 queue. In VALUETOOLS 2008, October 2008.Google ScholarDigital Library
- Samuli Aalto, Urtzi Ayesta, and Rhonda Righter. On the Gittins index in the M/G/1 queue. Queueing 6 Performance Evaluation Review, Vol. 49, No. 4, March 2022 Systems, 63(1):437--458, 2009.Google Scholar
- Samuli Aalto, Urtzi Ayesta, and Rhonda Righter. Properties of the Gittins index with application to optimal scheduling. Probability in the Engineering and Informational Sciences, 25(3):269--288, 2011.Google ScholarDigital Library
- Eitan Bachmat, Josu Doncel, and Hagit Sarfati. Analysis of the task assignment based on guessing size policy. Performance Evaluation, 142, 2020.Google Scholar
- Nikhil Bansal and Mor Harchol-Balter. Analysis of SRPT scheduling: Investigating unfairness. In Proceedings of ACM SIGMETRICS, pages 279--290, Cambridge, MA, June 2001.Google ScholarDigital Library
- D.R. Cox and W.L. Smith. Queues. Kluwer Academic Publishers, 1971.Google ScholarCross Ref
- Mark Crovella, Mor Harchol-Balter, and Cristina Murta. Task assignment in a distributed system: Improving performance by unbalancing load. In Proceedings of the ACM SIGMETRICS Joint International Conference on Measurement and Modeling of Computer Systems, pages 268--269, June 1998. Poster Session.Google Scholar
- Anshul Gandhi, Sherwin Doroudi, Mor Harchol-Balter, and Alan Scheller-Wolf. Exact analysis of the M/M/k/setup class of Markov chains via Recursive Renewal Reward. In ACM SIGMETRICS 2013 Conference on Measurement and Modeling of Computer Systems, pages 153--166, 2013.Google ScholarDigital Library
- Anshul Gandhi and Mor Harchol-Balter. How data center size impacts the effectiveness of dynamic power management. In 49th Annual Allerton Conference on Communication, Control, and Computing, pages 1164--1169, Urbana-Champaign, IL, September 2011.Google ScholarCross Ref
- John C. Gittins, Kevin D. Glazebrook, and Richard Weber. Multi-armed Bandit Allocation Indices. John Wiley & Sons, 2011.Google ScholarCross Ref
- Isaac Grosof, Ziv Scully, and Mor Harchol-Balter. SRPT for multiserver systems. Performance Evaluation, 127:154--175, November 2018.Google ScholarCross Ref
- Mor Harchol-Balter. Task assignment with unknown duration. Journal of the ACM, 49(2):260--288, March 2002.Google ScholarDigital Library
- Mor Harchol-Balter. Queueing disciplines. Wiley Encyclopedia of Operations Research and Management Science, 2011.Google ScholarCross Ref
- Mor Harchol-Balter. Performance Modeling and Design of Computer Systems: Queueing Theory in Action. Cambridge University Press, 2013.Google ScholarCross Ref
- Mor Harchol-Balter, Mark Crovella, and Cristina Murta. On choosing a task assignment policy for a distributed server system. In Lecture Notes in Computer Science, No. 1469: 10th International Conference on Modeling Techniques and Tools for Computer Performance Evaluation, pages 231--242, September 1998.Google Scholar
- Mor Harchol-Balter, Alan Scheller-Wolf, and Andrew Young. Surprising results on task assignment in server farms with high-variability workloads. In ACM SIGMETRICS 2009 Conference on Measurement and Modeling of Computer Systems, pages 287--0298, 2009.Google ScholarDigital Library
- Mor Harchol-Balter, Alan Scheller-Wolf, and Andrew Young. Why segregating short jobs from long jobs under high variability is not always a win. In Forty-seventh Annual Allerton Conference on Communication, Control, and Computing, pages 121--127, University of Illinois at Urbana-Champaign, October 2009.Google ScholarDigital Library
- Mor Harchol-Balter, Bianca Schroeder, Nikhil Bansal, and Mukesh Agrawal. Size-based scheduling to improve web performance. ACM Transactions on Computer Systems, 21(2):207--233, May 2003.Google ScholarDigital Library
- Mor Harchol-Balter and Rein Vesilo. To balance or unbalance load in size-interval task allocation. Probability in the Engineering and Informational Sciences, 24:219--244, 2010.Google ScholarDigital Library
- J.F.C. Kingman. Two similar queues in parallel. Biometrika, 48:1316--1323, 1961.Google Scholar
- Behnam Montazeri, Yilong Li, Mohammad Alizadeh, and John K. Ousterhout. Homa: A receiver-driven low-latency transport protocol using network priorities. In Proceedings of SIGCOMM 2018, pages 221--235. ACM SIGCOMM, 2018.Google ScholarDigital Library
- Alan Scheller-Wolf and Rein Vesilo. Structural interpretation and derivation of necessary and sufficient conditions for delay moments in fifo multiserver queues. Queueing Systems, 54(3):221--232, 2006.Google ScholarDigital Library
- Bianca Schroeder, Adam Wierman, and Mor Harchol-Balter. Open versus closed: a cautionary tale. In Proceedings of Networked Systems Design and Implementation (NSDI), 2006.Google Scholar
- Ziv Scully, Isaac Grosof, and Mor Harchol-Balter. The Gittins policy is nearly optimal in the M/G/k under extremely general conditions. Proceedings of ACM on Measurement and Analysis of Computer Systems (POMACS/SIGMETRICS), 4(3):1--29, 2020. Article 43.Google ScholarDigital Library
- Ziv Scully, Isaac Grosof, and Mor Harchol-Balter. Optimal multiserver scheduling with unknown job sizes in heavy traffic. In 38th International Symposium on Computer Performance, Modeling, Measurement, and Evaluation (IFIP PERFORMANCE 2020), Milan, Italy, November 2020.Google ScholarDigital Library
- Ziv Scully and Mor Harchol-Balter. The Gittins policy in the M/G/1 queue. In 19th International Symposium on Modeling and Optimization in Mobile, Ad hoc, and Wireless Networks (WiOpt '21), Philadelphia, PA, October 2021.Google ScholarCross Ref
- Ziv Scully and Mor Harchol-Balter. How to schedule near-optimally under real-world constraints. arXiv, 2021.Google Scholar
- Ziv Scully, Mor Harchol-Balter, and Alan Scheller-Wolf. SOAP: One clean analysis of all age-based scheduling policies. Proceedings of ACM on Measurement and Analysis of Computer Systems (POMACS/SIGMETRICS), 2(1):1--30, 2018. Article 16.Google Scholar
- Muhammad Tirmazi, Adam Barker, Nan Deng, MD E. Haque, Zhijing Gene Qin, Steven Hand, Mor Harchol-Balter, and John Wilkes. Borg: The next generation. In Proceedings of the Fifteenth European Conference on Computer Systems (EuroSys '20), pages 1--14, Greece, April 2020.Google ScholarDigital Library
- Adam Wierman and Mor Harchol-Balter. Classifying scheduling policies with respect to unfairness in an M/GI/1. In Proceedings of ACM SIGMETRICS, pages 238--249, San Diego, CA, June 2003.Google ScholarDigital Library
Index Terms
- The most common queueing theory questions asked by computer systems practitioners
Recommendations
SRPT optimally utilizes faster machines to minimize flow time
We analyze the shortest remaining processing time (SRPT) algorithm with respect to the problem of scheduling n jobs with release times on m identical machines to minimize total flow time. It is known that SRPT is optimal if m = 1 but that SRPT has a ...
SRPT Scheduling Discipline in Many-Server Queues with Impatient Customers
The shortest-remaining-processing-time (SRPT) scheduling policy has been extensively studied, for more than 50 years, in single-server queues with infinitely patient jobs. Yet, much less is known about its performance in multiserver queues. In this paper, ...
Single machine common flow allowance scheduling with a rate-modifying activity
In this paper we consider single machine SLK due date assignment scheduling problem with a rate-modifying activity. In this model, the machine has a rate-modifying activity that can change the processing rate of machine under consideration. Hence the ...
Comments