The Annals of Applied Probability

Dynamic scheduling of a system with two parallel servers in heavy traffic with resource pooling: asymptotic optimality of a threshold policy

S. L. Bell and R. J. Williams

Full-text: Open access


This paper concerns a dynamic scheduling problem for a queueing system that has two streams of arrivals to infinite capacity buffers and two (nonidentical) servers working in parallel. One server can only process jobs from one buffer. The service time distribution may depend on the buffer being served and the server providing the service. The system manager dynamically schedules waiting jobs onto available servers. We consider a parameter regime in which the system satisfies both a heavy traffic condition and a resource pooling condition. Our cost function is a mean cumulative discounted cost of holding jobs in the system, where the (undiscounted) cost per unit time is a linear function of normalized (with heavy traffic scaling) queue length. We first review the analytic solution of the Brownian control problem (formal heavy traffic approximation) for this system. We "interpret" this solution by proposing a threshold control policy for use in the original parallel server system. We show that this policy is asymptotically optimal in the heavy traffic limit and the limiting cost is the same as the optimal cost in the Brownian control problem. The techniques developed here are expected to be useful for analyzing the performance of threshold-type policies in more complex multiserver systems.

Article information

Ann. Appl. Probab., Volume 11, Number 3 (2001), 608-649.

First available in Project Euclid: 5 March 2002

Permanent link to this document

Digital Object Identifier

Mathematical Reviews number (MathSciNet)

Zentralblatt MATH identifier

Primary: 60K25: Queueing theory [See also 68M20, 90B22] 68M20: Performance evaluation; queueing; scheduling [See also 60K25, 90Bxx] 90B22: Queues and service [See also 60K25, 68M20] 90B35: Scheduling theory, deterministic [See also 68M20]
Secondary: 60J70: Applications of Brownian motions and diffusion theory (population genetics, absorption problems, etc.) [See also 92Dxx]

Queueing networks dynamic control resource pooling heavy traffic


Bell, S. L.; Williams, R. J. Dynamic scheduling of a system with two parallel servers in heavy traffic with resource pooling: asymptotic optimality of a threshold policy. Ann. Appl. Probab. 11 (2001), no. 3, 608--649. doi:10.1214/aoap/1015345343.

Export citation


  • [1] Billingsley, P. (1968). Convergence of Probability Measures. Wiley, New York.
  • [2] Bramson, M. (1996). Convergence to equilibria for fluid models of FIFO queueing networks. Queueing Systems 22 5-45.
  • [3] Chen, H. and Mandelbaum, A. (1991). LeontiefSystems, RBV's and RBM's. In Applied Stochastic Analysis (M. H. A. Davis and R. J. Elliott, eds.) 1-43. Gordon and Breach, New York.
  • [4] Chevalier, P. B. and Wein, L. (1993). Scheduling networks ofqueues: heavy traffic analysis ofa multistation closed network. Operations Research 41 743-758.
  • [5] Dembo, A. and Zeitouni, O. (1998). Large Deviations Techniques and Applications, 2nd ed. Springer, New York.
  • [6] Ethier, S. N. and Kurtz, T. G. (1986). Markov Processes: Characterization and Convergence. Wiley, New York.
  • [7] Harrison, J. M. (1985). Brownian Motion and Stochastic Flow Systems. Wiley, New York.
  • [8] Harrison, J. M. (1988). Brownian models ofqueueing networks with heterogeneous customer populations. In Stochastic Differential Systems, Stochastic Control Theory and Their Applications (W. Fleming and P. L. Lions, eds.) 147-186. Springer, New York.
  • [9] Harrison, J. M. (1996). The BIGSTEP approach to flow management in stochastic processing networks. In Stochastic Networks: Theory and Applications (F. P. Kelly, S. Zachary and I. Ziedins, eds.) 57-90. Oxford Univ. Press.
  • [10] Harrison, J. M. (1998). Heavy traffic analysis of a system with parallel servers: asymptotic optimality ofdiscrete-review policies. Ann. Appl. Probab. 8 822-848.
  • [11] Harrison, J. M. (2000). Brownian models ofopen processing networks: canonical representation ofworkload. Ann. Appl. Probab. 10 75-103.
  • [12] Harrison, J. M. and L ´opez, M. J. (1999). Heavy traffic resource pooling in parallel-server systems. Queueing Systems 33 339-368.
  • [13] Harrison, J. M. and Van Mieghem, J. A. (1997). Dynamic control ofBrownian networks: state space collapse and equivalent workload formulations. Ann. Appl. Probab. 7 747- 771.
  • [14] Harrison, J. M. and Wein, L. (1989). Scheduling networks ofqueues: heavy traffic analysis ofa simple open network. Queueing Systems 5 265-280.
  • [15] Harrison, J. M. and Wein, L. (1990). Scheduling networks ofqueues: heavy traffic analysis ofa two-station closed network. Oper. Res. 38 1052-1064.
  • [16] Iglehart, D. L. and Whitt, W. (1971). The equivalence offunctional central limit theorems for counting processes and associated partial sums. Ann. Math. Statist. 42 1372-1378.
  • [17] Jordan, W. C. and Graves, C. (1995). Principles on the benefits ofmanufacturing process flexibility. Management Sci. 41 577-594.
  • [18] Kelly, F. P. and Laws, C. N. (1993). Dynamic routing in open queueing networks: Brownian models, cut constraints and resource pooling. Queueing Systems 13 47-86.
  • [19] Kumar, S. (2000). Two-server closed networks in heavy traffic: diffusion limits and asymptotic optimality. Ann. Appl. Probab. 10 930-961.
  • [20] Kushner, H. J. and Chen, Y. N. (2000). Optimal control ofassignment ofjobs to processors under heavy traffic. Stochastics Stochastics Rep. 68 177-228.
  • [21] Kushner, H. J. and Dupuis, P. (1992). Numerical Methods for Stochastic Control Problems in Continuous Time. Springer, New York.
  • [22] Kushner, H. J. and Martins, L. F. (1990). Routing and singular control for queueing networks in heavy traffic. SIAM J. Control Optim. 28 1209-1233.
  • [23] Kushner, H. J. and Martins, L. F. (1996). Heavy traffic analysis of a controlled multiclass queueing network via weak convergence methods. SIAM J. Control Optim. 34 1781-1797.
  • [24] Laws, C. N. (1992). Resource pooling in queueing networks with dynamic routing. Adv. Appl. Probab. 24 699-726.
  • [25] Laws, C. N. and Louth, G. M. (1990). Dynamic scheduling ofa four-station queueing network. Probab. Engrg. Inform. Sci. 4 131-156.
  • [26] Martins, L. F., Shreve, S. E. and Soner, H. M. (1996). Heavy traffic convergence of a controlled, multi-class queueing system. SIAM J. Control Optim. 34 2133-2171.
  • [27] Mitrani, I. (1998). Probabilistic Modelling. Cambridge Univ. Press.
  • [28] Peterson, W. P. (1991). A heavy traffic limit theorem for networks of queues with multiple customer types. Math. Oper. Res. 16 90-118.
  • [29] Puhalskii, A. A. and Reiman, M. I. (1998). A critically loaded multirate link with trunk reservation. Queueing Systems 28 157-190.
  • [30] Rockafellar, R. T. (1970). Convex Analysis. Princeton Univ. Press.
  • [31] Roughan, M. and Pearce, C. E. M. (2000). A martingale analysis ofhysteretic overload control. Advances in Performance Analysis 3 1-30.
  • [32] Teh, Y. C. (1999). Threshold routing strategies for queueing networks. D.Phil. thesis, Univ. Oxford.
  • [33] Van Mieghem, J. A. (1995). Dynamic scheduling with convex delay costs: the generalized cµ rule. Ann. Appl. Probab. 5 808-833.
  • [34] Wein, L. (1990). Scheduling networks ofqueues: heavy traffic analysis ofa two-station network with controllable inputs. Oper. Res. 38 1065-1078.
  • [35] Williams, R. J. (1998). An invariance principle for semimartingale reflecting Brownian motions in an orthant. Queueing Systems 30 5-25.
  • [36] Williams, R. J. (1998). Diffusion approximations for open multiclass queueing networks: sufficient conditions involving state space collapse. Queueing Systems 30 27-88.
  • [37] Williams, R. J. (2000). On dynamic scheduling ofa parallel server system with complete resource pooling. In Analysis of Communication Networks: Call Centres, Traffic and Performance (D. R. McDonald and S. R. E. Turner, eds.) 49-71. Amer. Math. Soc., Providence, RI.