Load Balancing Model Research Articles

Load-balancing evaluation model based on pre-defined variability

An evaluation model of load-balancing was presented for finite difference parallel computing.Maximal Load Variability(MLV) was introduced as a key index for load-balancing,and the quantitative model was constructed with parallel efficiency as the object function,involving ratio of communication to computing,degree of parallel,problem size and complexity of local numerical schemes.Both the parameter MLV and the model were verified by analyzing the POP global ocean circulation model benchmark.The results show that the performance derived from the evaluation model is consistent with that of wall-time measurement on the whole.And the model given in this paper is also useful for those parallel applications with local schemes such as finite element and finite volume.

A bipartite model for load balancing in grid computing environments

In this paper, a bipartite model for load balancing (LB) in grid computing environments, called Transverse viewpoint-based Bi-Tier model (TBT), is proposed. TBT can efficiently eliminate topology mismatching between overlay- and physical-networks during the load transfer process. As an implementation of TBT, a novel LB policy called M2ON (Min-cost and Max-flow Channel based Overlay Network) is presented. In M2ON, the communication capability is denoted as M2C (Min-cost and Max-flow Channel) which is obtained using a Labeled Tree Probing (LTP) method. The computing capacity is denoted as the Idle Factor (IF) which is obtained from the semantic overlay. The higher- and lower-level characteristics are combined into an Integrated Impacting Factor (IIF) using a Double Linear Inserting (DLI) function. Based on IIF, optimal topology matching can be achieved in the LB process. Extensive experiments and simulations have been performed and will be discussed. The results show that M2ON achieves more accurate topology matching with a minimum increment in the overall locating time yet achieving higher system performance as a whole.

A repartitioning hypergraph model for dynamic load balancing

In parallel adaptive applications, the computational structure of the applications changes over time, leading to load imbalances even though the initial load distributions were balanced. To restore balance and to keep communication volume low in further iterations of the applications, dynamic load balancing (repartitioning) of the changed computational structure is required. Repartitioning differs from static load balancing (partitioning) due to the additional requirement of minimizing migration cost to move data from an existing partition to a new partition. In this paper, we present a novel repartitioning hypergraph model for dynamic load balancing that accounts for both communication volume in the application and migration cost to move data, in order to minimize the overall cost. The use of a hypergraph-based model allows us to accurately model communication costs rather than approximate them with graph-based models. We show that the new model can be realized using hypergraph partitioning with fixed vertices and describe our parallel multilevel implementation within the Zoltan load balancing toolkit. To the best of our knowledge, this is the first implementation for dynamic load balancing based on hypergraph partitioning. To demonstrate the effectiveness of our approach, we conducted experiments on a Linux cluster with 1024 processors. The results show that, in terms of reducing total cost, our new model compares favorably to the graph-based dynamic load balancing approaches, and multilevel approaches improve the repartitioning quality significantly.

Data load balance model for multi-cluster grid

To improve the utilization efficiency of heterogeneous resources in Multi-Cluster Grid (MCG) composed of many computer clusters, a data load balance model was proposed. According to computation power, storage power, and communication power of query node, the performance model of query node was studied. By using data saturation, matrix of data load balance and data migration technology, system's data load balance mechanism was described. Experiment results show:The model can be fit for querying massive data.

Wind integration into various generation mixtures

A load balance model is used to quantify the economic and environmental effects of integrating wind power into three typical generation mixtures. System operating costs over a specified period are minimized by controlling the operating schedule of the existing power generating facilities for a range of wind penetrations. Unlike other studies, variable generator efficiencies, and thus variable fuel costs, are taken into account, as are the ramping constraints on thermal generators. Results indicate that the system operating cost will increase by 83%–280% (pending generation mixture) at a wind penetration of 100% of peak demand. System emissions also decrease by 13%–32% (depending on the generation mixture) at a wind penetration of 100%. This leads to emission abatement costs in the range of $1300/tonne-CO 2e for hydro dominated mixtures, $240/tonne-CO 2e for coal dominated mixtures, and $215/tonne-CO 2e for natural gas dominated mixtures.

Web server load balancing: A queueing analysis

Over the last few years, the Web-based services, more specifically different types of E-Commerce applications, have become quite popular, resulting in exponential growth in the Web traffic. In many situations, this has led to unacceptable response times and unavailability of services, thereby driving away customers. Many companies are trying to address this problem using multiple Web servers with a front-end load balancer. Load balancing has been found to provide an effective and scalable way of managing the ever-increasing Web traffic. However, there has been little attempt to analyze the performance characteristics of a system that uses a load balancer. This paper presents a queuing model for analyzing load balancing with two Web servers. We first analyze the centralized load balancing model, derive the average response time and the rejection rate, and compare three different routing policies at the load balancer. We then extend our analysis to the distributed load balancing and find the optimal routing policy that minimizes the average response time.

Fuzzy logic based neural network models for load balancing in wireless networks

In this paper, adaptive channel borrowing approach fuzzy neural networks for load balancing (ACB-FNN) is presented to maximized the number of served calls and the depending on asymmetries traffic load problem. In a wireless network, the call's arrival rate, the call duration and the communication overhead between the base station and the mobile switch center are vague and uncertain. A new load balancing algorithm with cell involved negotiation is also presented in this paper. The ACB-FNN exhibits better learning abilities, optimization abilities, robustness, and fault-tolerant capability thus yielding better performance compared with other algorithms. It aims to efficiently satisfy their diverse quality-of-service (QoS) requirements. The results show that our algorithm has lower blocking rate, lower dropping rate, less update overhead, and shorter channel acquisition delay than previous methods.

Stability of Load Balancing Algorithms in Dynamic Adversarial Systems

In the dynamic load balancing problem, we seek to keep the job load roughly evenly distributed among the processors of a given network. The arrival and departure of jobs is modeled by an adversary restricted in its power. Muthukrishnan and Rajaraman [An adversarial model for distributed dynamic load balancing, in Proceedings of the 10th ACM Symposium on Parallel Algorithms and Architectures, ACM, New York, 1998] gave a clean characterization of a restriction on the adversary that can be considered the natural analogue of a cut condition. They proved that a simple local balancing algorithm proposed by Aiello et al. [Approximate load balancing on dynamic and asynchronous networks, in Proceedings of the 25th ACM Symposium on Theory of Computing, ACM, New York, 1993] is stable against such an adversary if the insertion rate is restricted to a $(1-\varepsilon)$ fraction of the cut size. They left as an open question whether the algorithm is stable at rate 1. In this paper, we resolve this question positively, by proving stability of the local algorithm at rate 1. Our proof techniques are very different from the ones used by Muthukrishnan and Rajaraman and yield a simpler proof and tighter bounds on the difference in loads. In addition, we introduce a multicommodity version of this load balancing model and show how to extend the result to the case of balancing two different kinds of loads at once (obtaining as a corollary a new proof of the 2-commodity Max-Flow Min-Cut Theorem). We also show how to apply the proof techniques to the problem of routing packets in adversarial systems. Awerbuch et al. [Simple routing strategies for adversarial systems, in Proceedings of the 42nd IEEE Symposium on Foundations of Computer Science, IEEE Computer Society, Los Alamitos, CA, 2001] showed that the same load balancing algorithm is stable against an adversary, inserting packets at rate 1 with a single destination in dynamically changing networks. Our techniques give a much simpler proof for a different model of adversarially changing networks.

Convergence time to Nash equilibrium in load balancing

We study the number of steps required to reach a pure Nash equilibrium in a load balancing scenario where each job behaves selfishly and attempts to migrate to a machine which will minimize its cost. We consider a variety of load balancing models, including identical, restricted, related, and unrelated machines. Our results have a crucial dependence on the weights assigned to jobs. We consider arbitrary weights, integer weights, k distinct weights, and identical (unit) weights. We look both at an arbitrary schedule (where the only restriction is that a job migrates to a machine which lowers its cost) and specific efficient schedulers (e.g., allowing the largest weight job to move first). A by-product of our results is establishing a connection between various scheduling models and the game-theoretic notion of potential games. We show that load balancing in unrelated machines is a generalized ordinal potential game, load balancing in related machines is a weighted potential game, and load balancing in related machines and unit weight jobs is an exact potential game.

The hierarchical model for load balancing on two machines

Following previous work, we consider the hierarchical load balancing model on two machines of possibly different speeds. We first focus on maximizing the minimum machine load and show that no competitive algorithm exists for this problem. We overcome this barrier in two ways, both related to previously known models. The first one is fractional assignment, where each job can be arbitrarily split between the machines. The second one is a semi-online model where the sum of jobs is known in advance. We design algorithms of best possible competitive ratios for both these cases. Furthermore, we show that the combination of the two models leads to the existence of an optimal algorithm (i.e., an algorithm of competitive ratio 1). This algorithm is clearly optimal for the makespan minimization problem as well. For the latter problem, we consider the fractional assignment model and design an algorithm of best possible competitive ratio for it.

Research of a Load Balancing Model Based on Mobile Agent

Research of a Load Balancing Model Based on Mobile Agent

Flow-Level Multipath Load Balancing in MPLS Network

Multi-Protocol Label Switching (MPLS) can efficiently support the explicit routes setup by the use of Label Switched Paths (LSPs) between an ingress Label Switched Router (LSR) and an egress LSR. Hence it is possible to distribute the network traffic among several paths to achieve load balancing, thus improving the network utilization, and minimizing the congestion. The packet-level traffic characteristics in the Internet is so complex that it is natural to do traffic engineering (TE) and control at the flow level. The emerging Multi-Protocol Label Switching (MPLS) has introduced an attractive solution to TE in IP networks. The main objective of this paper is to balance traffic at the flow level among the parallel Label Switched Paths (LSPs) in MPLS networks. We introduce a multipath load-balancing model at the flow level. In this model, each LSP is modeled as an M/G/1 processor-sharing queue. The load-balancing problem is then considered as an optimization problem. Based on the analysis of the model, we propose a heuristic but efficient mechanism that can make good use of the traffic characteristics at the flow level. Packet disorder is avoided effectively by dispatching packets belonging to one flow to the same path. This mechanism only need to be implemented in the ingress LSRs and the egress LSRs, while the intermediate LSRs only forward the packets. Apart from discussing the traffic allocation granularity, and the implementation issues in details, we have also performed extensive simulations using NS-2 with MPLS modules. The simulation results show that the load through the network is well balanced so that the network throughput is improved and the delay is decreased efficiently.

A three-stage load balancing model in a manufacturing company

We present a three-stage model aimed at balancing the production system of a company manufacturing pants. The company is characterized by having is main force-spot in the ability of the workmen at making a determinate task within the production system. In particular, the human factor and the random events are really crucial for the company in order to be competitive: the former because of the capability of people at making a specific work, the latter because of the possibility that a workman, for unpredictable reasons, can be absent and thus the current distribution of the work might result not well balanced. This means that, at any time, it might be necessary to re-balance the production system, taking into account not only the capabilities of people, but also the absence of some workmen and the need to substitute them. This problem has been treated by a three-stage optimization model, involving linear programming and open-shop scheduling.

A Dynamic Load Balancing Model for a Distributed System

A communication protocol designed for fault tolerance in distributed real time systems is implemented and a dynamic load balancing model is designed and implemented over this protocol. The protocol consists of cluster based, hierarchical rings which use synchronous communication. The rings are synchronous. At the lowest level in the hierarchy, there are clusters that consist of computing processors, called nodes. The higher level consists of the cluster representatives that manage the clusters of the lower level. There can be two or more levels in the hierarchy. Ring protocols in each cluster can work in parallel. Also, a fault tolerance mechanism is integrated to the protocol. The dynamic distributed load balancing module designed over the protocol aims to maximize the overall performance of the whole system by distributing the load submitted to the system efficiently and transparently among the nodes. While performing operations to achieve this goal, the module also considers the real-time constraints of the system

Dynamic Load Balancing for Short-range Parallel Molecular Dynamics Simulations

The iterative Multilevel Averaging Weight (MAW) algorithm presented in paper [1] is modified to solve the dynamic load imbalance problems arising from the two-dimensional short-range parallel molecular dynamics simulations in this paper. Firstly, five types of load balancing models are given which allows detailed studies of the algorithm. In particular, it shows that for strip decomposition, the number of iteration needs for the system to converge from an initially unbalanced state to a well balanced state is bounded by 2 log P , where P is the number of processors. This result can permit the algorithm to efficiently track fluctuations in the molecular density as the simulation progresses, and is much better than that of the Cellular Automaton Diffusion (CAD) scheme presented in paper [2] . Secondly, we apply MAW algorithm to solve the load imbalance problem in the parallel molecular dynamics simulation for higher speed wall collisions. At last, the numerical experimental results and parallel computing performance with MPI-1.2 under a PC-Cluster consists of 64 Pentium-III 500 MHz nodes connected by 100 Mbps Switches are given in this paper.

A general matrix iterative model for dynamic load balancing

Many applications require some form of load balancing in order to obtain good performance on parallel computing systems. A number of load balancing algorithms have been proposed, and implemented on real systems. However, application developers are in need of a performance tool that compares and evaluates various balancing algorithms, and helps select the best algorithm for a given application and computing platform. In this paper, we propose a matrix iterative model that can represent a wide range of dynamic load balancing algorithms. The model captures the change in task distribution over the processors due to both the application imbalance and load balancing algorithm. We also define performance measures that capture the costs of executing the parallel application and load balancing algorithm on a given computing platform. The model and the performance measures are used to compare and evaluate various load balancing algorithms. In this paper, the model is parameterized to represent three load balancing algorithms — the random, diffusion and redistribution algorithms. Model predictions and experimental measurements are compared for the parallel N-body simulation application. This application is representative of a wide class of scientific applications where time-varying workload distributions are present. The model predictions are within 20% of measured values. The performance of the three algorithms are compared, and optimal algorithm parameters are derived showing how the model can be used to both select an algorithm and set its parameters.

Multithreaded model for the dynamic load-balancing of parallel adaptive PDE computations

We present a multithreaded model for the dynamic load-balancing of numerical, adaptive computations required for the solution of Partial Differential Equations (PDEs) on multicomputers. Multithreading is used as a means of exploring concurrency in the processor level in order to tolerate synchronization costs inherent to traditional (nonthreaded) parallel adaptive PDE solvers. Our preliminary analysis for parallel, adaptive PDE solvers indicates that multithreading can be used as a mechanism to mask overheads required for the dynamic balancing of processor workloads with computations required for the actual numerical solution of the PDEs. Also, multithreading can simplify the implementation of dynamic load-balancing algorithms, a task that is very difficult for traditional data-parallel adaptive PDE computations. Unfortunately, multithreading does not always simplify program complexity, often makes code reusability not an easy task, and increases software complexity.

A parallel simulation model for load balancing in clustered distributed systems

The paper analyses the features of clustered distributed systems and proposes a parallel model for their simulation. This model is used for the implementation of a parallel simulator on a shared memory parallel computer. Same adapted variants of classical static and dynamic scheduling algorithms are then tested using the simulator. In the end, the results obtained using these load balancing strategies are compared with those obtained in the absence of load balancing.

A network flow model for load balancing in circuit-switched multicomputers

In multicomputers that utilize circuit switching or wormhole routing, communication overhead depends largely on link contention-the variation due to distance between nodes is negligible. This has a major impact on the load balancing problem. In this case there are some nodes with an excess load (sources) and other with a deficit load (sinks). A matching of sources to sinks is required to avoid contention. The problem is made complex by the hardwired routing on currently available machines: The user can control only which nodes communicate but not how the messages are routed. Network flow models of message flow in the mesh and the hypercube have been developed to solve this problem. The crucial property of these models is the correspondence between minimum cost flows and correctly routed messages. To solve a given load balancing problem, a minimum cost flow algorithm is applied to the network. This permits the efficient determination of a maximum contention free matching of sources to sinks that, in turn, tells how much of the given imbalance can be eliminated without contention.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Queueing network models for load balancing in distributed systems

In distributed systems, load balancing can improve efficiency by migrating jobs from heavily loaded to lightly loaded sites. In this paper we present a method for optimal load allocation in a static environment. A queueing network model is used to evaluate response time, and mathematical programming techniques are used to find the load allocation that minimizes average response time. The method is not proposed as a substitute for dynamic, heuristic load balance policies; rather, it is perceived as a useful tool for resource allocation and capacity planning in distributed systems and as a promising complement to dynamic policies in hybrid load balance strategies. The method can handle very general classes of problems, including distinct classes of jobs, multitasking within each job, and jobs with spawned tasks. Several examples illustrating these applications are reported.