Stationary Waiting Time Distribution Research Articles

Stationary Waiting Time in Parallel Queues with Synchronization

Motivated by database locking problems in today’s massive computing systems, we analyze a queueing network with many servers in parallel (files) to which jobs (writing access requests) arrive according to a Poisson process. Each job requests simultaneous access to a random number of files in the database and will lock them for a random period of time. Alternatively, one can think of a queueing system where jobs are split into several fragments that are then randomly routed to specific servers in the network to be served in a synchronized fashion. We assume that the system operates on a first-come, first-served basis. The synchronization and service discipline create blocking and idleness among the servers, which leads to a strict stability condition compared with other distributed queueing models. We analyze the stationary waiting time distribution of jobs under a many-server limit and provide exact tail asymptotics. These asymptotics generalize the celebrated Cramér–Lundberg approximation for the single-server queue.

An M/M/1 Queueing-Inventory System with Working Vacations, Vacation Interruptions and Lost Sales

We consider a single server queueing-inventory system under (s,Q) replenishment strategy with working vacations, vacation interruptions and lost sales. The server provides service at a lower rate during working vacations than while in normal mode of service. If a working vacation realizes while providing service in that mode, then the server continues in the present status until the current service is completed. Upon that it switches to the normal mode of service, provided there is at least one customer waiting. If no customer is waiting at this point of time, it goes for vacation. We also assume that if there are customers in the system at a service completion epoch during a working vacation, the server will come back to the normal working mode, otherwise the server will stay in the working vacation mode. With the system having infinite capacity, the stability condition for the system is obtained, followed by computation of the steady-state probability vector and discussion. Various performance measures are evaluated. In addition, the busy period analysis is provided and the stationary waiting time distribution in the queue is derived. Numerical illustrations are provided to illustrate the system performance, and an optimization problem is also discussed.

Delay performance of data-center queue with setup policy and abandonment

This paper considers a finite capacity multiserver queue for performance modeling of data centers. One of the most important issues in data centers is to save energy of servers. To this end, a natural policy is to turn off a server immediately once it has no job to process. In this policy, which is called ON–OFF policy, the server must be setup upon the arrival of a new job. During the setup time, the server cannot process jobs but consumes energy. To mitigate the drawback, this paper considers a setup policy, where the number of setup servers at a time is limited. We also consider an extension of the setup policy in which some of servers, but not all of them, are allowed to remain idle for a random amount of time. The main purpose of this paper is to analyze the delay distribution of the multiserver queue with the setup policies. We assume that jobs may abandon the queue without receiving their services. We formulate the queue length process using a two-dimensional continuous-time Markov chain. It can be shown that we cannot apply the distributional Little’s law to obtain the waiting time distribution via the queue length distribution. Therefore, we construct a three-dimensional absorbing Markov chain which describes the virtual waiting time process. We then obtain a phase-type expression of the stationary waiting time distribution. We evaluate the performance of the multiserver queue with the setup policy, and then discuss the optimal control parameters of the policy based on the delay performance as well as the mean power consumption.

Extension of the loss probability formula to an overloaded queue with impatient customers

We consider a batch arrival MX∕G∕1 queue with impatient customers. The loss probability is expressed in terms of the stationary waiting time distribution for the standard MX∕G∕1 queue with no impatience. But this expression is only applicable when the offered load ρ is less than 1. We give a formula for the loss probability applicable for any values of ρ>0, by proving that the loss probability is analytic in ρ on (0,∞) through a Girsanov-type change of measure.

The MX/M/1 queue with working breakdown

In this paper, we consider a batch arrival MX/M/1 queue model with working breakdown. The server may be subject to a service breakdown when it is busy, rather than completely stoping service, it will decrease its service rate. For this model, we analyze a two-dimensional Markov chain and give its quasi upper triangle transition probability matrix. Under the system stability condition, we derive the probability generating function (PGF) of the stationary queue length, and then obtain its stochastic decomposition, which shows the relationship with that of the classical MX/M/ 1 queue without vacation or breakdown. Besides, we also give the Laplace-Stieltjes transform (LST) of the stationary waiting time distribution of an arbitrary customer in a batch. Finally, some numerical examples are given to illustrate the effect of the parameters on the system performance measures.

The M<sup>X</sup>/M/1 Queue with Multiple Working Vacation

We study a batch arrival MX/M/1 queue with multiple working vacation. The server serves customers at a lower rate rather than completely stopping service during the service period. Using a quasi upper triangular transition probability matrix of two-dimensional Markov chain and matrix analytic method, the probability generating function (PGF) of the stationary system length distribution is obtained, from which we obtain the stochastic decomposition structure of system length which indicates the relationship with that of the MX/M/1 queue without vacation. Some performance indices are derived by using the PGF of the stationary system length distribution. It is important that we obtain the Laplace Stieltjes transform (LST) of the stationary waiting time distribution. Further, we obtain the mean system length and the mean waiting time. Finally, numerical results for some special cases are presented to show the effects of system parameters.

On temporal characteristics in an exponential queueing system with negative claims and a bunker for ousted claims

We consider a queueing system with Poisson input streams of positive and negative claims, an infinite collector, and exponential service. A negative claim ousts a positive claim out of the collector queue and moves it to a bunker of unbounded capacity. If the collector is empty then a negative claim leaves the system with no influence on it. After a claim is serviced, the device receives the next claim from the collector or, if the collector is empty, from the bunker. For different combinations of FIFO and LIFO orders of choosing a claim for service from the collector's queue, choosing a claim for service from the bunker's queue, and ousting claims from the collector to the bunker, we obtain formulas for computing the stationary waiting time distribution for a claim to begin service and other temporal characteristics.

Algorithm for waiting time distribution of a discrete-time multiserver queue with deterministic service times and multi-threshold service policy

In this paper, a discrete-time multiserver queueing system with infinite buffer size is studied. Packets arrive at the system according to a Bernoulli arrival process, and the service times of the packets are assumed to be constant equal to an integer multiple of slots. The number of active servers in the system is controlled by a so-called multi-threshold service policy with a set of thresholds predefined. For this multiserver system, we derive its stationary waiting time distribution by using the stochastic complement and Crommelin's techniques. We also present an algorithm for computing the stationary waiting time distribution.

Equivalences of Batch-Service Queues and Multi-Server Queues and Their Complete Simple Solutions in Terms of Roots

In this article, we first present a unified discussion of several equivalence relationships among (as well as between) batch-service queues and multi-server queues, in terms of the stationary queue-length and waiting-time distributions. Then, we present a complete and simple solution for the queue-length and waiting-time distributions of the discrete-time multi-server deterministic-service Geo/D/b queue, in terms of roots of the so-called characteristic equation. This solution also represents the solutions for the other equivalent queues, as a result of the equivalence relationships. To aid in the applications of these results, sample numerical results are presented at the end.

A Single-Server Queueing System with Background Customers

A single-server queueing system with a Markov flow of primary customers and a flow of background customers from a bunker containing unbounded number of customers, i.e., the background customer flow is saturated, is studied. There is a buffer of finite capacity for primary customers. Service processes of primary as well as background customers are Markovian. Primary customers have a relative service priority over background customers, i.e., a background customer is taken for service only if the buffer is empty upon completion of service of a primary customer. A matrix algorithm for computing the stationary state probabilities of the system at arbitrary instants and at instants of arrival and completion of service of primary customers is obtained. Main stationary performance indexes of the system are derived. The Laplace--Stieltjes transform of the stationary waiting time distribution for primary customers is determined.

Relaxation Time for the Discrete D/G/1 Queue

When queueing models are used for performance analysis of some stochastic system, it is usually assumed that the system is in steady-state. Whether or not this is a realistic assumption depends on the speed at which the system tends to its steady-state. A characterization of this speed is known in the queueing literature as relaxation time. The discrete D/G/1 queue has a wide range of applications. We derive relaxation time asymptotics for the discrete D/G/1 queue in a purely analytical way, mostly relying on the saddle point method. We present a simple and useful approximate upper bound which is sharp in case the load on the system is not very high. A sharpening of this upper bound, which involves the complementary error function, is then developed and this covers both the cases of low and high loads. For the discrete D/G/1 queue, the stationary waiting time distribution can be expressed in terms of infinite series that follow from Spitzer's identity. These series involve convolutions of the probability distribution of a discrete random variable, which makes them suitable for computation. For practical purposes, though, the infinite series should be truncated. The relaxation time asymptotics can be applied to determine an appropriate truncation level based on a sharp estimate of the error caused by truncating.

COMPUTATIONAL ANALYSIS OF STATIONARY WAITING-TIME DISTRIBUTIONS OF GIX/R/1 AND GIX/D/1 QUEUES

In this article, we obtain, in a unified way, a closed-form analytic expression, in terms of roots of the so-called characteristic equation of the stationary waiting-time distribution for the GIX/R/1 queue, where R denotes the class of distributions whose Laplace–Stieltjes transforms are rational functions (ratios of a polynomial of degree at most n to a polynomial of degree n). The analysis is not restricted to generalized distributions with phases such as Coxian-n (Cn) but also covers nonphase-type distributions such as deterministic (D). In the latter case, we get approximate results. Numerical results are presented only for (1) the first two moments of waiting time and (2) the probability that waiting time is zero. It is expected that the results obtained from the present study should prove to be useful not only for practitioners but also for queuing theorists who would like to test the accuracies of inequalities, bounds, or approximations.

M/GI/1 queues with services of both positive and negative customers

We consider an M/GI/1 queue with two types of customers, positive and negative, which cancel each other out. The server provides service to either a positive customer or a negative customer. In such a system, the queue length can be either positive or negative and an arrival either joins the queue, if it is of the same sign, or instantaneously removes a customer of the opposite sign at the end of the queue or in service. This study is a generalization of Gelenbe's original concept of a queue with negative customers, where only positive customers need services and negative customers arriving at an empty system are lost or need no service. In this paper, we derive the transient and the stationary probability distributions for the major performance measures in terms of generating functions and Laplace transforms. It has been shown that the previous results for the system with negative arrivals of zero service time are special cases of our model. In addition, we obtain the stationary waiting time distribution of this system in terms of a Laplace transform.

M/GI/1 queues with services of both positive and negative customers

We consider an M/GI/1 queue with two types of customers, positive and negative, which cancel each other out. The server provides service to either a positive customer or a negative customer. In such a system, the queue length can be either positive or negative and an arrival either joins the queue, if it is of the same sign, or instantaneously removes a customer of the opposite sign at the end of the queue or in service. This study is a generalization of Gelenbe's original concept of a queue with negative customers, where only positive customers need services and negative customers arriving at an empty system are lost or need no service. In this paper, we derive the transient and the stationary probability distributions for the major performance measures in terms of generating functions and Laplace transforms. It has been shown that the previous results for the system with negative arrivals of zero service time are special cases of our model. In addition, we obtain the stationary waiting time distribution of this system in terms of a Laplace transform.

Some Useful Functions for Functional Large Deviations

Useful descriptions of stochastic models are often provided when they are represented as functions of well understood stochastic models. Properties of the well understood model can be preserved by the representation. For example, the large deviation principle (LDP) is preserved by continuous maps via the contraction principle and weak convergence is preserved by continuous maps via the continuous mapping theorem. In a recent paper, Ganesh and O'Connell demonstrate the value of particular topological space of functions indexed by the positive real line by showing: (1) a large class of partial sums process satisfy the functional LDP in the space; and (2) the supremum map is continuous when restricted to an appropriate subspace. This makes the space useful, for example, in proving logarithmic asymptotics for the stationary waiting-time distribution at a single server queue. This paper facilitates further deductions from the space by considering other useful maps: inversion, composition and first passage time. It is shown that composition and inversion are continuous when restricted to an appropriate subspace. First passage time is not continuous, but with additional rate-function assumptions the LDP can be deduced by explicit calculation. A number of examples of these results are presented.

AN INFINITE-PHASE QUASI-BIRTH-AND-DEATH MODEL FOR THE NON-PREEMPTIVE PRIORITY M/PH/1 QUEUE

This paper considers a single server queue that handles arrivals from N classes of customers on a non-preemptive priority basis. Each of the N classes of customers features arrivals from a Poisson process at rate λ i and class-dependent phase type service. To analyze the queue length and waiting time processes of this queue, we derive a matrix geometric solution for the stationary distribution of the underlying Markov chain. A defining characteristic of the paper is the fact that the number of distinct states represented within the sub-level is countably infinite, rather than finite as is usually assumed. Among the results we obtain in the two-priority case are tractable algorithms for the computation of both the joint distribution for the number of customers present and the marginal distribution of low-priority customers, and an explicit solution for the marginal distribution of the number of high-priority customers. This explicit solution can be expressed completely in terms of the arrival rates and parameters of the two service time distributions. These results are followed by algorithms for the stationary waiting time distributions for high- and low-priority customers. We then address the case of an arbitrary number of priority classes, which we solve by relating it to an equivalent three-priority queue. Numerical examples are also presented.

An M/ G/1 queue with Markov-dependent exceptional service times

This paper considers an M/ G/1 queue in which service time distributions in each busy period change according to a finite state Markov chain, embedded at the arrival instants of customers. It is assumed that this Markov chain has an upper triangular transition matrix. Applying the regenerative cycle approach with respect to a busy period, we obtain the Laplace–Stieltjes transform, i.e., LST, of the stationary waiting time distribution in a certain parametric form. We give a procedure to determine those parameters. Some detailed calculations and numerical results are presented as well.

Distributional Form of Little's Law for FIFO Queues with Multiple Markovian Arrival Streams and Its Application to Queues with Vacations

This paper considers stationary queues with multiple arrival streams governed by an irreducible Markov chain. In a very general setting, we first show an invariance relationship between the time-average joint queue length distribution and the customer-average joint queue length distribution at departures. Based on this invariance relationship, we provide a distributional form of Little's law for FIFO queues with simple arrivals (i.e., the superposed arrival process has the orderliness property). Note that this law relates the time-average joint queue length distribution with the stationary sojourn time distributions of customers from respective arrival streams. As an application of the law, we consider two variants of FIFO queues with vacations, where the service time distribution of customers from each arrival stream is assumed to be general and service time distributions of customers may be different for different arrival streams. For each queue, the stationary waiting time distribution of customers from each arrival stream is first examined, and then applying the Little's law, we obtain an equation which the probability generating function of the joint queue length distribution satisfies. Further, based on this equation, we provide a way to construct a numerically feasible recursion to compute the joint queue length distribution.

A REGENERATIVE CYCLE APPROACH TO AN M/G/1 QUEUE WITH EXCEPTIONAL SERVICE

We are concerned with an M/G/1 queue in which service time distributions in each busy period may depend on the number of ~ust~omers who have been served in the same busy period. This model is called an exceptional service model. Our major interest is t,o see a general structure of this model through the stationary waiting time distribution and some other ~haracterist~ics. To t,his end, we take a regenerative cycle approach with respect to a busy cycle. This approach enables us to get several characteristics in tractable forms, from which we get some interesting; properties of the exceptional service model. Numerical examples are presented as well.

A heavy‐traffic theorem for the GI/G/1 queue with a Pareto‐type service time distribution

For the GI/G/1 queueing model with traffic load a < 1, service time distribution B(t) and interarrival time distribution A(t), whenever for t → ∞ urn:x-wiley:20903332:media:ista149413:ista149413-math-0001 , and urn:x-wiley:20903332:media:ista149413:ista149413-math-0002 , converges in distribution for a↑1. Here w is distributed as the stationary waiting time distribution. The L.‐S. transform of the limiting distribution is derived and an asymptotic series for its tail probabilities is obtained. The theorem actually proved in the text concerns a slightly more general asymptotic behavior of 1 − B(t), t → ∞, than mentioned above.