Periodic load balancing

Multiprocessor load balancing aims to improve performance by moving jobs from highly loaded processors to more lightly loaded processors. Some schemes allow only migration of new jobs upon arrival, while other schemes allow migration of jobs in progress. A difficulty with all these schemes, however, is that they require continuously maintaining detailed state information. In this paper we consider the alternative of periodic load balancing, in which the loads are balanced only at each T time units for some appropriate T. With periodic load balancing, state information is only needed at the balancing times. Moreover, it is often possible to use slightly stale information collected during the interval between balancing times. In this paper we study the performance of periodic load balancing. We consider multiple queues in parallel with unlimited waiting space to which jobs come either in separate independent streams or by assignment (either random or cyclic) from a single stream. Resource sharing is achieved by periodically redistributing the jobs or the work in the system among the queues. The performance of these systems of queues coupled by periodic load balancing depends on the transient behavior of a single queue. We focus on useful approximations obtained by considering a large number of homogeneous queues and a heavy load. When the number of queues is sufficiently large, the number of jobs or quantity of work at each queue immediately after redistribution tends to evolve deterministically, by the law of large numbers. The steady-state (limiting) value of this deterministic sequence is obtained as the solution of a fixed point equation, where the initial value is equal to the expected transient value over the interval between successive redistributions conditional on the initial value. A refined approximation based on the central limit theorem is a normal distribution, where the mean and variance are obtained by solving a pair of fixed-point equations. With higher loads, which is natural to consider when load balancing is performed, a heavy-traffic limit theorem shows that one-dimensional reflected Brownian motion can be used to approximately describe system performance, even with general arrival and service processes. With these approximations, we show how performance depends on the assumed arrival pattern of jobs and the model parameters. We do numerical calculations and conduct simulation experiments to show the accuracy of the approximations.

[1]  Feller William,et al.  An Introduction To Probability Theory And Its Applications , 1950 .

[2]  S. Zhou,et al.  A Trace-Driven Simulation Study of Dynamic Load Balancing , 1987, IEEE Trans. Software Eng..

[3]  Aleksandr Alekseevich Borovkov,et al.  Stochastic processes in queueing theory , 1976 .

[4]  Ward Whitt,et al.  Transient behavior of regulated Brownian motion, II: Non-zero initial conditions , 1987 .

[5]  Anthony P. Reeves,et al.  Strategies for Dynamic Load Balancing on Highly Parallel Computers , 1993, IEEE Trans. Parallel Distributed Syst..

[6]  K. K. Ramakrishnan,et al.  UNITE-an architecture for lightweight signaling in ATM networks , 1998, Proceedings. IEEE INFOCOM '98, the Conference on Computer Communications. Seventeenth Annual Joint Conference of the IEEE Computer and Communications Societies. Gateway to the 21st Century (Cat. No.98.

[7]  Ward Whitt,et al.  Calculating time-dependent performance measures for the M/M/1 queue , 1989, IEEE Trans. Commun..

[8]  Ward Whitt,et al.  Dependence in packet queues , 1989, IEEE Trans. Commun..

[9]  G. J. Foschini,et al.  Unintrusive communication of status in a packet network in heavy traffic , 1985, AT&T Technical Journal.

[10]  R. Weber On the optimal assignment of customers to parallel servers , 1978, Journal of Applied Probability.

[11]  François Baccelli,et al.  Elements Of Queueing Theory , 1994 .

[12]  Richard F. Serfozo,et al.  Partitions of point processes: Multivariate poisson approximations , 1985 .

[13]  W. Whitt,et al.  Transient behavior of regulated Brownian motion, I: Starting at the origin , 1987, Advances in Applied Probability.

[14]  Ward Whitt,et al.  Peak congestion in multi-server service systems with slowly varying arrival rates , 1997, Queueing Syst. Theory Appl..

[15]  Gísli Hjálmtýsson Lightweight call setup — Supporting connection and connectionless services , 1997 .

[16]  Michael H. Rothkopf,et al.  Perspectives on Queues: Combining Queues is Not Always Beneficial , 1987, Oper. Res..

[17]  N. L. Lawrie,et al.  Comparison Methods for Queues and Other Stochastic Models , 1984 .

[18]  Tapani Lehtonen,et al.  On the optimality of the shortest line discipline , 1984 .

[19]  W. Whitt,et al.  Resource sharing for efficiency in traffic systems , 1981, The Bell System Technical Journal.

[20]  Ward Whitt,et al.  Sensitivity to the Service-Time Distribution in the Nonstationary Erlang Loss Model , 1995 .

[21]  C. Laws Resource pooling in queueing networks with dynamic routing , 1992, Advances in Applied Probability.

[22]  Ward Whitt,et al.  Understanding the Efficiency of Multi-Server Service Systems , 1992 .

[23]  W. Whitt Weak convergence theorems for priority queues: preemptive-resume discipline , 1971, Journal of Applied Probability.

[24]  Ward Whitt,et al.  Numerical Inversion of Laplace Transforms of Probability Distributions , 1995, INFORMS J. Comput..

[25]  Martin I. Reiman,et al.  Some diffusion approximations with state space collapse , 1984 .

[26]  Richard L. Tweedie,et al.  Markov Chains and Stochastic Stability , 1993, Communications and Control Engineering Series.

[27]  Zhang Hanqin,et al.  MULTIPLE CHANNEL QUEUES IN HEAVY TRAFFIC , 1990 .

[28]  W. Whitt,et al.  Diffusion approximations for queues with server vacations , 1990, Advances in Applied Probability.

[29]  William Feller,et al.  An Introduction to Probability Theory and Its Applications , 1967 .

[30]  Ward Whitt,et al.  Deciding Which Queue to Join: Some Counterexamples , 1986, Oper. Res..

[31]  Ward Whitt,et al.  The Fourier-series method for inverting transforms of probability distributions , 1992, Queueing Syst. Theory Appl..

[32]  P. Billingsley,et al.  Convergence of Probability Measures , 1969 .

[33]  Amnon Barak,et al.  The MOSIX Distributed Operating System: Load Balancing for UNIX , 1993 .

[34]  Hirotaka Sakasegawa,et al.  An approximation formulaLq ≃α·ρβ/(1-ρ) , 1977 .

[35]  W. Whitt Comparing counting processes and queues , 1981, Advances in Applied Probability.

[36]  Bruce E. Hajek,et al.  Performance of global load balancing of local adjustment , 1990, IEEE Trans. Inf. Theory.

[37]  W. Whitt,et al.  Transient behavior of the M/M/1 queue via Laplace transforms , 1988, Advances in Applied Probability.

[38]  Teunis J. Ott,et al.  Load-balancing heuristics and process behavior , 1986, SIGMETRICS '86/PERFORMANCE '86.

[39]  E. A. Doorn Stochastic Monotonicity and Queueing Applications of Birth-Death Processes , 1981 .

[40]  Avishai Mandelbaum,et al.  On Pooling in Queueing Networks , 1998 .

[41]  Irene A. Stegun,et al.  Handbook of Mathematical Functions. , 1966 .

[42]  G. F. Newell,et al.  Introduction to the Theory of Queues. , 1963 .

[43]  R. Wolff AN UPPER BOUND FOR MULTI-CHANNEL QUEUES , 1977 .

[44]  Milton Abramowitz,et al.  Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables , 1964 .

[45]  M. R. Taaffe,et al.  Approximating nonstationaryPh(t)/M(t)/s/c queueing systems , 1987 .

[46]  Ward Whitt,et al.  Some Useful Functions for Functional Limit Theorems , 1980, Math. Oper. Res..

[47]  D. Iglehart,et al.  Multiple channel queues in heavy traffic. I , 1970, Advances in Applied Probability.

[48]  W. Whitt Planning queueing simulations , 1989 .

[49]  Edward D. Lazowska,et al.  Adaptive load sharing in homogeneous distributed systems , 1986, IEEE Transactions on Software Engineering.

[50]  W. Whitt,et al.  Multidimensional Transform Inversion with Applications to the Transient M/G/1 Queue , 1994 .

[51]  D. Iglehart,et al.  Multiple channel queues in heavy traffic. II: sequences, networks, and batches , 1970, Advances in Applied Probability.

[52]  Ward Whitt,et al.  Comparison methods for queues and other stochastic models , 1986 .

[54]  G. J. Foschini,et al.  A Basic Dynamic Routing Problem and Diffusion , 1978, IEEE Trans. Commun..

[55]  Rongxin Wang,et al.  Heavy traffic limit theorems for a sequence of shortest queueing systems , 1995, Queueing Syst. Theory Appl..