Abstract
Heavy-traffic delay optimality is considered to be an important metric in evaluating the delay performance of load balancing schemes. In this paper, we argue that heavy-traffic delay optimality is a coarse metric that does not necessarily imply good delay performance. Specifically, we show that any load balancing scheme is heavy-traffic delay optimal as long as it satisfies a fairly weak condition. This condition only requires that in the long-term the dispatcher favors, even slightly, shorter queues over longer queues. Hence, although a load balancing scheme could be heavy-traffic delay optimal, the empirical delay performance of heavy-traffic delay optimal schemes can range from very good (that of join-shortest-queue) to very bad (arbitrarily close to the performance of random routing). To overcome this limitation, we introduce a new metric called degree of queue imbalance, which measures the queue length difference between all the servers in steady-state. Given a heavy-traffic delay optimal load balancing scheme, we can characterize the resultant degree of queue imbalance. This, in turn, allows us to explicitly differentiate between good and poor load balancing schemes. Thus, this paper suggests that when designing good load balancing schemes, they should not only be heavy-traffic delay optimal, but also have a low degree of queue imbalance.
- Dimitris Bertsimas, David Gamarnik, and John N Tsitsiklis. 2001. Performance of multiclass Markovian queueing networks via piecewise linear Lyapunov functions. Annals of Applied Probability (2001), 1384--1428.Google Scholar
- Maury Bramson. 1998. State space collapse with application to heavy traffic limits for multiclass queueing networks. Queueing Systems, Vol. 30, 1--2 (1998), 89--140. Google ScholarDigital Library
- Mark E Crovella, Mor Harchol-Balter, and Cristina D Murta. 1998. Task assignment in a distributed system (extended abstract): improving performance by unbalancing load. In ACM SIGMETRICS Performance Evaluation Review, Vol. Vol. 26. ACM, 268--269. Google ScholarDigital Library
- Atilla Eryilmaz and R Srikant. 2012. Asymptotically tight steady-state queue length bounds implied by drift conditions. Queueing Systems, Vol. 72, 3--4 (2012), 311--359. Google ScholarDigital Library
- G Foschini and JACK Salz. 1978. A basic dynamic routing problem and diffusion. IEEE Transactions on Communications Vol. 26, 3 (1978), 320--327.Google ScholarCross Ref
- Ian Foster, Yong Zhao, Ioan Raicu, and Shiyong Lu. 2008. Cloud computing and grid computing 360-degree compared 2008 Grid Computing Environments Workshop. Ieee, 1--10.Google Scholar
- David Gamarnik, John N Tsitsiklis, and Martin Zubeldia. 2016. Delay, Memory, and Messaging Tradeoffs in Distributed Service Systems Proceedings of the 2016 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Science. ACM, 1--12. Google ScholarDigital Library
- Lars George. 2011. HBase: the definitive guide. " O'Reilly Media, Inc.".Google Scholar
- Varun Gupta, Mor Harchol Balter, Karl Sigman, and Ward Whitt. 2007. Analysis of join-the-shortest-queue routing for web server farms. Performance Evaluation Vol. 64, 9 (2007), 1062--1081. Google ScholarDigital Library
- FP Kelly and CN Laws. 1993. Dynamic routing in open queueing networks: Brownian models, cut constraints and resource pooling. Queueing systems, Vol. 13, 1 (1993), 47--86.Google Scholar
- Yi Lu, Qiaomin Xie, Gabriel Kliot, Alan Geller, James R Larus, and Albert Greenberg. 2011. Join-Idle-Queue: A novel load balancing algorithm for dynamically scalable web services. Performance Evaluation Vol. 68, 11 (2011), 1056--1071. Google ScholarDigital Library
- Siva Theja Maguluri, R Srikant, et almbox.. 2016. Heavy traffic queue length behavior in a switch under the MaxWeight algorithm. Stochastic Systems, Vol. 6, 1 (2016), 211--250.Google ScholarCross Ref
- Siva Theja Maguluri, R Srikant, and Lei Ying. 2014. Heavy traffic optimal resource allocation algorithms for cloud computing clusters. Performance Evaluation Vol. 81 (2014), 20--39. Google ScholarDigital Library
- Michael Mitzenmacher. 2001. The power of two choices in randomized load balancing. IEEE Transactions on Parallel and Distributed Systems, Vol. 12, 10 (2001), 1094--1104. Google ScholarDigital Library
- Michael Mitzenmacher. 2016. Analyzing distributed Join-Idle-Queue: A fluid limit approach Communication, Control, and Computing (Allerton), 2016 54th Annual Allerton Conference on. IEEE, 312--318.Google ScholarCross Ref
- Debankur Mukherjee, Sem C Borst, Johan SH van Leeuwaarden, and Philip A Whiting. 2016. Universality of Power-of-$ d $ Load Balancing in Many-Server Systems. arXiv preprint arXiv:1612.00723 (2016).Google Scholar
- Martin I Reiman. 1984. Some diffusion approximations with state space collapse. Modelling and performance evaluation methodology. Springer, 207--240.Google Scholar
- Jori Selen, Ivo Adan, Stella Kapodistria, and Johan van Leeuwaarden. 2016. Steady-state analysis of shortest expected delay routing. Queueing Systems, Vol. 84, 3--4 (2016), 309--354. Google ScholarDigital Library
- Alexander L Stolyar. 2015. Pull-based load distribution in large-scale heterogeneous service systems. Queueing Systems, Vol. 80, 4 (2015), 341--361. Google ScholarDigital Library
- Alexander L Stolyar et almbox.. 2004. Maxweight scheduling in a generalized switch: State space collapse and workload minimization in heavy traffic. The Annals of Applied Probability Vol. 14, 1 (2004), 1--53.Google ScholarCross Ref
- Weina Wang, Kai Zhu, Lei Ying, Jian Tan, and Li Zhang. 2016. MapTask scheduling in MapReduce with data locality: Throughput and heavy-traffic optimality. IEEE/ACM Transactions on Networking Vol. 24, 1 (2016), 190--203. Google ScholarDigital Library
- Ruth J Williams. 1998. Diffusion approximations for open multiclass queueing networks: sufficient conditions involving state space collapse. Queueing systems, Vol. 30, 1 (1998), 27--88. Google ScholarDigital Library
- Qiaomin Xie and Yi Lu. 2015. Priority algorithm for near-data scheduling: Throughput and heavy-traffic optimality Computer Communications (INFOCOM), 2015 IEEE Conference on. IEEE, 963--972.Google Scholar
- Qiaomin Xie, Ali Yekkehkhany, and Yi Lu. 2016. Scheduling with multi-level data locality: Throughput and heavy-traffic optimality Computer Communications, IEEE INFOCOM 2016-The 35th Annual IEEE International Conference on. IEEE, 1--9.Google ScholarCross Ref
Index Terms
- Degree of Queue Imbalance: Overcoming the Limitation of Heavy-traffic Delay Optimality in Load Balancing Systems
Recommendations
Designing Low-Complexity Heavy-Traffic Delay-Optimal Load Balancing Schemes: Theory to Algorithms
SIGMETRICS '18: Abstracts of the 2018 ACM International Conference on Measurement and Modeling of Computer SystemsWe establish a unified analytical framework for designing load balancing algorithms that can simultaneously achieve low latency, low complexity, and low communication overhead. We first propose a general class ¶ of load balancing policies and prove that ...
Heavy-traffic Delay Optimality in Pull-based Load Balancing Systems: Necessary and Sufficient Conditions
In this paper, we consider a load balancing system under a general pull-based policy. In particular, each arrival is randomly dispatched to one of the servers with queue length below a threshold; if none exists, this arrival is randomly dispatched to ...
Designing Low-Complexity Heavy-Traffic Delay-Optimal Load Balancing Schemes: Theory to Algorithms
In this paper, we establish a unified analytical framework for designing load balancing algorithms that can simultaneously achieve low latency, low complexity, and low communication overhead. We first propose a general class \Pi of load balancing ...
Comments