Decentralized Dynamic Load Balancing: The Particles Approach

Size: px
Start display at page:

Download "Decentralized Dynamic Load Balancing: The Particles Approach"

Transcription

1 appered in: Information Sciences, Vol 84, Issue 1-2 (May 1995) S Decentralized Dynamic Load Balancing: The Particles Approach Hans-Ulrich Heiss Department of Informatics and Automation Technical University of Ilmenau, Germany Michael Schmitz Department of Informatics University of Karlsruhe, Germany Abstract: We consider the problem of mapping tasks to processor nodes at run-time in multiprogrammed multicomputer systems (ie message passing MIMD-systems) Besides load balancing, the goal is to place intensively communicating tasks close together to minimize communication delays Since the placement has to be done at run-time by migrating tasks, migration cost is also considered Our decentralized approach is inspired by a physical analogy: Tasks are considered as particles acted upon by forces Each aspect of the allocation goal is modeled by a dedicated force Migration activities cease, when a stable situation with balanced forces is reached Simulations for multiprogrammed hypercube and 2D-grid topologies confirmed the usefulness and suggests superiority of this more general approach to other decentralized algorithms for the environment considered

2 1 INTRODUCTION Multicomputers are MIMD-machines with local, private memories and an interconnection network for message passing We assume that each processor is autonomous in the sense that it is controlled by a private operating system kernel which provides basic functions such as task generation, multitasking, and intertask communication (within the node and across node boundaries) In addition, we consider multiprogramming operation, ie many independent parallel programs may run at the same time, sharing the machine in time and space The programs may enter the system at any node and at any time Each of these programs is assumed to consist of a dynamic set of parallel tasks that communicate with each other by interchanging messages In such an environment, there exists the problem of how to (re)allocate the particular tasks to processors to achieve low response times Two classes of approaches can be distinguished: Static Mapping (contractive embedding of parallel programs) Given a parallel program with m communicating tasks and a multicomputer with n<m processors, the problem is to find a mapping of the tasks to the processors that minimizes the program's execution time Heuristic algorithms for this NP-hard problem try to find a balanced allocation with minimal communication delays They are often based on graph partitioning techniques [16] and use modern optimization heuristics such as Simulated Annealing or Genetic Algorithms [5,14] However, they are designed to be executed off-line (eg in a separate step at link time running on the front-end of the parallel machine) and consider neither the actual load situation (due to multiprogramming) at run-time nor the data dependent behavior of the program They are usually also unable to deal with dynamic task creation and deletion during program execution Dynamic Load Balancing The other class of approaches originates from distributed systems where new tasks may enter the system at any time and at any node The problem is to distribute the tasks as evenly as possible to avoid idling nodes If, in addition, the tasks are elements of parallel programs, load balancing helps to process tasks at different 2

3 processors at the same speed, which leads to low synchronization delays and thus to low program response times Since the allocation is done at run-time, the basic mechanism is the migration of a task from one node to another which usually means the transfer of a considerable amount of data Most of the proposals use a decentralized algorithm where load balancing agents at the particular nodes negotiate the possible transfer of tasks from overloaded to underloaded nodes Intertask communication is usually ignored and if considered, it plays a minor role, since the algorithms are mostly dedicated to systems based on local area networks where all processors can be regarded as adjacent [7,10,18] For the considered environment as described above, a task allocation mechanism needs a combination of features and goals of both classes We therefore propose a decentralized dynamic load balancing mechanism that explicitly considers communication and migration overhead Our algorithm is based on a physical model that uses the notion of forces that correspond to independent optimization goals The basic idea behind our approach is the following: Imagine a flat container with an even bottom where different amounts of non mixable fluids of different viscosity are placed at different points (Figure 1) Gravitation forces the fluids to run but frictional resistance and cohesion forces that make up the viscosity are working counter Thinner fluids may spread out evenly across the bottom of the container, while more viscous fluids stick together like a lump After some time, the fluid distribution will reach a stable state with balanced forces Using this image we consider the parallel computations as fluids with the tasks as particles The load potential at each node in the system can be used to define a gravitational force Communication relations along with their intensities are associated with the cohesion forces in direction and magnitude Costs to migrate tasks act as frictional resistance and are also working counter to load balancing By using this model, our algorithm pursues the following goals in parallel: (i) (ii) (iii) (iv) Minimization of load unbalance Minimization of communication costs Avoidance of unproductive migrations Stability, ie avoidance of oscillations 3

4 The rest of the paper is organized as follows: In section 2, we introduce our system model and define the particular forces which are used to explain the algorithm in section 3 Section 4 presents some results from simulations conducted for a 2D-grid machine topology, and section 5 concludes the paper summarizing the major features of our approach 2 THE SYSTEM MODEL A natural formal model of a multicomputer is a processor connection graph PCG = (P,L) with P set of processor elements (vertices) L P P set of interprocessor links (edges) a(k,l) time it takes to transport one data unit from processor k to processor l µ k speed of processor k (# instructions per second) A single parallel program consists of a set of interacting tasks or threads and can therefore also be modeled as a graph, known as task interaction graph TIG = (T,C) with T set of tasks (vertices) C T T set of communication channels (edges) c(i,j) communication intensity, (i, j) C (# data units) s i d i length of task t i (# instructions to be executed) size of description of task t i (amount of data to be migrated) Programs are generated dynamically with random interarrival times The generation (or arrival) site is chosen randomly 21 Load Balancing Force Forces are the result of different levels of potentials A load balancing force is therefore based on the definition of a load potential In the framework of our model, three different definitions of the load potential of a node are possible: (i) the number of tasks assigned k V load := { j: loc( j ) = k} (1a) (ii) the amount of work assigned k V load := s j j: loc( j)=k (1b) 4

5 k 1 (iii) the time it takes to execute the assigned work V load := s µ j (1c) k j:loc( j)=k where loc(i) denotes the index of the processor node where task t i is located Definition (iii) is most accurate because it considers different processor speeds and different sizes of the tasks If all processors are identical, then definition (ii) is equivalent If all tasks are alike or if we do not know the lengths of the tasks we must resort to definition (i) Principally, any other quantity expressing the load-related delay can be used For the results reported below, we assume definition (1a) Based on this load potential one could define a load balancing force between two adjacent nodes as proportional to the difference of the two potentials However, since load balancing is intended to achieve roughly equal processing delay at different nodes, a relative measure of the load difference is more appropriate than an absolute one Assume, for instance, a load potential using (1a) and a local round-robin scheduling policy and consider two pairs of nodes, one pair with load potentials of 1 and 5, the other pair with 11 and 15: Both pairs have the same load difference of 4 but with regard to the processing speed this means in the former case a factor of 5, while in the latter a factor of only 15/11 Leveling out the load difference at the former processor pair is therefore more important Thus, we prefer to define as load balancing force the ratio of the potentials: j f k lb (t i ) := c lb V load +1 j k ( )( V load +1 ) (2) as the magnitude of a force acting on a task t i which is located at node j pointing to node k, where j and k are direct neighbors The increment +1 in (2) is applied simply to deal with potentials of 0 The coefficient c lb can be considered a universal constant used for weighting the different forces 22 Communication Force (Cohesion) The communication force is based on the communication intensities between the tasks To calculate it, we need to know, at least approximately, the current position of the communication partners The communication cost for a pair of communicating tasks can be defined as the product of the amount of data sent and the distance Assuming that the communication intensity c(i,j) is zero for non-communicating tasks, we define a 5

6 communication potential for a task t i residing at node k as its total communication costs (totalled over all communication partners): k V com T (t i ) := a(k,loc(j)) c(i, j) (3) j =1 The force trying to minimize communication cost for task t i is then defined as the difference of the potentials: j k j (t i ) := c com V com (t i ) V k com (t i ) f com ( ) (4) j In other words, f k com (t i ) indicates the gain in communication cost incurred by t i by moving t i from node j to node k 23 Damping Forces Migration means that the current state of the task has to be shipped which in the worst case is its entire address space Shipment and installing the task at the target node means cost, communication cost and execution cost as well Thus, migration is useful only in cases where the gain achieved by migration outweighs the incurred cost This is analogous to the physical friction of a body which acts as a counterforce to any other force attracting the body Any attracting force must exceed the friction in magnitude to actually move the body Consequently, we model friction as migration costs Let d i be the description of task t i, ie the amount of data that must be transferred for a migration Then the frictional force is defined as j k f frict (ti ) := c frict d i a( j,k) (5) Again, c frict serves as a constant to adjust the combined effect of the different forces In addition to friction, we introduce another damping component that counteracts migration: With each task t i, we associate a migration counter z i that is initialized to zero when the task is generated, and incremented each time the task is migrated A constant called max migs defines the maximum number of migrations permitted A migration takes place only if x i = 1 with z 1, if r i x i : = max migs (6) 0, otherwise 6

7 where r is a random variable drawn from a uniform distribution over the interval [0,1) Besides the fact that it reduces algorithm overhead by limiting the number of eligible tasks, it makes sure that no task will be sent around forever, since a task is migrated at most max migs times Therefore, after an initial placement of tasks and if no load is added or removed, migrations will finally cease and bring the load balancing activities to a stop 3 THE ALGORITHM Load balancing should take place when the load situation has changed The load situation changes when tasks are generated or finished Generating new tasks makes the generating site a possible sender of load It therefore initiates a load balancing step If tasks are finished, the finishing site informs all direct neighbors about the load change and continues normal operation If necessary, one or more of the informed neighbors start the load balancing procedure After collecting all required load information from its neighbors, the load balancing node determines which of its tasks are eligible for migration by evaluating the x i 's according to (6) For those tasks eligible, all forces are calculated that attract the task into one of the possible directions defined by the direct links The resultant is composed as a linear combination of the different force types: j k j k j k j k f res (ti ) := f lb (ti ) + f com (ti ) + f frict (ti ) (7) At node j this expression is the force component acting into direction of k Because migrations are only possible along the links, it makes no sense to calculate the total resultant force as it would be done in the Euclidean space Instead, we take the maximum of all these direction components and use it as the total force acting on a task (Figure 2) Of all tasks that are attracted to a neighbor, we choose that with the maximum magnitude and initiate its migration After each migration, the source node informs its neighbors about the new load situation, which may lead to a domino effect of load balancing activities At each node, the load balancing stops, if either no task is eligible for migration or all forces are negative (Figure 3) Considering the load balancing force only, those asynchronous, decentralized load balancing algorithms can be proven to converge to an equilibrium load distribution [2, 3] With the additional communication force, however, the algorithm realizes a 7

8 distributed gradient search which tends to converge into a local minimum if not the damping factors bring the algorithm to a premature stop However, since we are considering a highly dynamic multiprogramming environment with arriving and departing programs, the 'landscape' of the objective function is permanently changing Thus, every algorithm is deemed to trail a moving global optimum in a heuristic way The algorithm assumes the knowledge of some program characteristics that may not be readily available The most difficult data to obtain is the communication intensity between tasks which may be unknown in advance There are different ways to address this problem, eg: Over different executions of a program, the operating system can measure its behavior and provide an execution profile of the program The profile describes quantitatively the expected behavior of the program If no such profile is available, the algorithm can start with the assumption that communication intensities are uniform, ie assuming a task interaction graph with unit edge weights During the course of the program's execution the actual intensities may be measured and used by the algorithm 4 SIMULATION RESULTS The algorithm was evaluated by simulations using PASTE (Processor Allocation Strategy Test Environment), a testbed designed as a tool to analyze task allocation algorithms in parallel computers [11] The machine topology used was a 8x8-twodimensional grid (A 6D-hypercube was also simulated, but did not produce significantly different results) We assumed a processor speed of 200 Mips per node and a communication capacity of 10 Mbyte/sec for each link We used an open model with a program arrival rate large enough to produce a considerable load Programs arrived at randomly selected nodes and immediately started execution (no queuing) Each program consisted on the average of 12 interacting tasks with an exponentially distributed execution time with a mean of 50 sec (10 10 instructions) Each task had between 1 and 6 communication partners (node degree of the TIG) The communication was synchronous requiring the sender to wait until the data was received by the partner 8

9 (rendezvous) Multitasking at each node was implemented with a round-robin scheduling policy The task size (amount of data for migration) was uniformly distributed between 50 and 500 Kbyte Total simulation time was controlled by a confidence-level criterion which stopped, when the mean response time reached an accuracy of ±5% with 90% confidence It roughly corresponded to 1000 program generations Each point in the following graphs represents one simulation run To show the versatility of our approach, we focus on the influence of the weighting coefficients on different performance measures In general, the optimal coefficient values mainly depend on the characteristics of the underlying machine architecture and are rather insensitive to changes in the program behavior For the hardware parameters used we found empirically that the parameter set (c lb =08, c com =65, c frict =25, and max migs =15) was roughly optimal with regard to the mean response time This parameter set was the basis for the following figures The effect of switching off some forces - simply by setting the corresponding coefficient to 0 - on the mean response time is shown in Figure 4, and gives some possibility of comparison to other algorithms already proposed Ignoring both, communication and migration costs makes our algorithm similar to the Gradient Model [13] Ignoring migration cost only, roughly corresponds to the mapping algorithm proposed in [4] Figure 4 indicates that for low load levels pure load balancing is sufficient, because the predominant goal is to avoid idle processor nodes Due to the rendezvous-type communication, load balancing also helps to keep communication delays low: If all tasks of a program proceed with the same speed, synchronization delays will be minimized At higher utilization levels, however, where the probability for idle nodes is low, the potential gain that can be achieved by sending tasks to idle nodes decreases, and the interest shifts from load sharing to overhead avoidance Therefore, as we increase the arrival rate, pure load balancing delivers the worst results and it becomes necessary to consider communication and migration overhead For the high load region (λ > 025 in Figure 4), where it is profitable to take into account other aspects than load balancing, we show in Figure 5 the sensitivity of three 9

10 performance indices with regard to changing the values of the weighting coefficients Besides the mean response time of the programs, Figure 5 depicts: - (effective) utilization, defined as the accumulated user busy times (excluding overhead caused by the load balancing activities) of a processor divided by the total simulation time and averaged over all processors, - load unbalance index, defined as the variance of utilization over the different processors, averaged over the simulation time (based on 10 second sampling intervals) Figure 5a shows the effect of c lb and confirms the expectation that load balancing is necessary to achieve low response time and high utilization Interesting is the course of the response time, since it gives some evidence of the existence of a minimum which means that there is some optimal value for c lb : Giving load balancing too much weight compared to other goals will deteriorate the performance Figure 5b reports the variation of the communication (cohesion) force coefficient Here, too, the response time curve exhibits a minimum meaning that some effort to minimize communication overhead is advantageous to system performance (Note that for programs with asynchronous communication, the impact of the force on response time would be somewhat weaker since communication could be partially overlapped with computation With the synchronous communication considered here, there is only overlapping due to multitasking at the nodes) Not surprisingly, the communication overhead (as well as the migration overhead, not shown here) becomes lower with increasing communication force (high viscosity) Figure 5c clearly indicates that migration costs have to be taken into account Up to an optimal value of the coefficient, the response times drop continuously As for the communication force, too much weight for this frictional force, however, hampers migration and load balancing, and deteriorates performance The migrating overhead is monotonically decreasing and the load unbalance index increasing, as expected All response time curves (left column in Figure 5) exhibit a convex functional dependency on the coefficients with a clear minimum This underlines the necessity to include the corresponding forces The figures indicate that - for the environment under consideration, ie multiprogramming of parallel programs - (i) minimizing response times, (ii) maximizing utilization, and (iii) minimizing load unbalance are different 10

11 goals that must be well distinguished We can have perfect load balance and a high utilization but nevertheless bad response times, since the tasks of the programs may not be well mapped with regard to their communication relations All figures confirm what could be expected from the model, eg that load balance and utilization are monotonically increasing functions of the influence of a load balancing force, while the cohesion and frictional forces have an opposite effect Not shown is the effect of varying the maximum number of migrations There is also an optimal value with regard to the mean response time, but the sensitivity is rather low Extensive experiments suggest a value of the order of the diameter of the processor topology Further increasing max migs may still improve the load balance, but may deteriorate response times 5 CONCLUSION We proposed a new decentralized heuristic algorithm to the problem of dynamically distribute tasks of parallel programs in large multicomputer systems with multiprogramming In contrast to other load balancing schemes it reflects communication patterns among the tasks of the parallel programs So it also solves the mapping problem for parallel programs at run-time The approach is based on a physical analogy and comprises load balancing and minimization of communication and migration overhead Weighting coefficients to balance and fine-tune the effects of the subgoals can be tailored to pursue a large range of different objectives Experiments based on simulations for 2D-grid and hypercube topologies confirm the necessity to account for communication and migration overhead in order to achieve low response times Especially for high loads, load balancing alone is not sufficient and the consideration of the additional forces can reduce the response times by up to 50% To further prove its usefulness in real operation, the proposed algorithm will be implemented in our distributed operating system COSY [6] which currently runs on transputer-clusters 11

12 REFERENCES 1 Ahmad, I; Ghafoor, A: Semi-Distributed Load Balancing for Massively Parallel Microcomputer Systems" IEEE Trans on Software Engineering 17,10 (Oct 1991) pp Bertsekas, DP; Tsitsiklis,JN: Parallel and Distributed Computation Prentice Hall (1989) 3 Boillat, EJ: Load Balancing and Poisson equation in a graph Concurrency: Practice and Experience 2,4 (Dec 1990) pp Boillat, E J; Kropf, PG: "A Fast Distributed Mapping Algorithm" CONPAR 90 (Int Conf on Vector and Parallel Processing) pp Bultan,T; Aykanat,C: A New Mapping Heuristic Based on Mean Field Annealing Journal of Parallel and Distributed Computing 16 (1992) pp Butenuth,R: The Cosy-Kernel as an Example for Efficint Kernel Call Mechanisms on Transputers Proc 6th Transputer/occam Int Conference (June 16-17, 1994) Tokyo, Japan 7 Casavant, TL; Kuhl,JG: "A Taxonomy of Scheduling in General-Purpose Distributed Computing Systems" IEEE Trans on Software Engineering 14,2 (Feb 1988) pp Chow,YC; Kohler,WK: "Models for Dynamic Load Balancing in a Heterogeneous Multiprocessor System" IEEE Trans on Computers 28,5 (May 1979) pp Cybenko,G: "Dynamic Load Balancing for Distributed Memory Multiprocessors" Journal of Parallel and Distributed Computing 7 (1989) pp Eager,DL; Lazowska,ED; Zahorjan,J: "A Comparison of Receiver-Initiated and Sender-Initiated Adaptive Load Sharing" Performance Evaluation 6 (1986) pp Heiss,H-U; Payer, A: "PASTE: A Toool for Evaluation of Processor Allocation Strategies" Proc 6th Int Conf on Modelling Tools and Techn for Comp Perf Eval, Edinburgh, Sept 1992, pp Heiss,H-U; Schmitz,M: "Distributed Load Balancing Using a Physical Analogy" Internal Report No5/92, Dep of Informatics, University of Karlsruhe (May 1992) 13 Lin,FCH; Keller,RM: "The Gradient Model Load Balancing Method" IEEE Trans on Software Engineering 13,1 (Jan 1987) pp Mühlenbein,H; Gorges-Schleuter,M; Krämer,O: "New solutions to the mapping problem of parallel systems: The evolution approach" Parallel Computing 4(1987), pp Nicol,DM; Saltz,JH: "Dynamic Remapping of Parallel Computations with Varying Resource Demands" IEEE Trans on Computers 37,9 (Sept 1988) pp Sadayappan,P; Ercal,F; Ramanujam,J: "Cluster partitioning approaches to mapping parallel programs onto a hypercube" Parallel Computing, Vol 13 (1990), pp Shin,K; Chang,Y: "Load Sharing in Distributed Real-Time Systems with State Change Broadcasts" IEEE Trans on Computers 38,8 (Aug 1989) pp Theimer,MM; Lantz,KA: "Finding Idle Machines in a Workstation-Based Distributed Systems" IEEE Trans on Software Engineering 15,11 (Nov 1989) pp

13 (a) adding a new fluid (b) after reaching stability (balanced forces) Figure 1: Situation in a container with different viscous fluids p 1 5 f 1 res (ti ) p 4 5 f 4 res (ti ) t i 5 f 2 res (ti ) p 2 p 5 5 f 3 res (ti ) p 3 Figure 2: Forces acting on t i along the links The maximum force component points to p 4 Thus, t i will be migrated to p 4 13

14 1 cycle Periodic activation 2 wait for load change 3 if load increase 4 then 5 M { i T x i = 1} Determine migratable tasks 6 load_balanced false 7 while M and not load_balanced do 8 get information from neighbors 9 for all i M do For all migratable tasks 10 for all k neighbors do calculate forces to all possible 11 calculate f(i,k) directions 12 end for 13 end for 14 cand i with max i M, k neighbors 15 direct k with max k neighbors { f( i,k) } Select task with largest force { f( cand,k) } Direction of largest force 16 if f(cand,direct) > 0 If force positive, migrate 17 then candidate to the direction of 18 migrate(cand, direct) largest force 19 send new load data to neighbors Update and distribute load data 20 end then to neighbors 21 else 22 load_balanced true 23 end else 24 end while 25 end then 26 else Load decrease 27 send new load data to neighbors Update and distribute load data 28 end else 29 end cycle Figure 3: Outline of the load balancing algorithm 14

15 95 response time [s] c lb utilization [%] c lb (a) Variation of the load balancing coefficient 07 load unbalance c lb response time [s] utilization [%] load unbalance ccom 52 ccom (b) Variation of the communication coefficient ccom response time [s] utilization [%] load unbalance c f rict (c) Variation of the frictional coefficient 52 c f rict c frict Figure 5: Impact of varying the strength of particular forces on different performance measures 15

16 1000 response time [s] complete load balancing and friction load balancing and communication load balancing only Figure 4: Mean response time as a function of the total load for different allocation schemes 16

DECENTRALIZED LOAD BALANCING IN HETEROGENEOUS SYSTEMS USING DIFFUSION APPROACH

DECENTRALIZED LOAD BALANCING IN HETEROGENEOUS SYSTEMS USING DIFFUSION APPROACH DECENTRALIZED LOAD BALANCING IN HETEROGENEOUS SYSTEMS USING DIFFUSION APPROACH P.Neelakantan Department of Computer Science & Engineering, SVCET, Chittoor pneelakantan@rediffmail.com ABSTRACT The grid

More information

Load balancing in a heterogeneous computer system by self-organizing Kohonen network

Load balancing in a heterogeneous computer system by self-organizing Kohonen network Bull. Nov. Comp. Center, Comp. Science, 25 (2006), 69 74 c 2006 NCC Publisher Load balancing in a heterogeneous computer system by self-organizing Kohonen network Mikhail S. Tarkov, Yakov S. Bezrukov Abstract.

More information

A Comparison of General Approaches to Multiprocessor Scheduling

A Comparison of General Approaches to Multiprocessor Scheduling A Comparison of General Approaches to Multiprocessor Scheduling Jing-Chiou Liou AT&T Laboratories Middletown, NJ 0778, USA jing@jolt.mt.att.com Michael A. Palis Department of Computer Science Rutgers University

More information

Distributed Dynamic Load Balancing for Iterative-Stencil Applications

Distributed Dynamic Load Balancing for Iterative-Stencil Applications Distributed Dynamic Load Balancing for Iterative-Stencil Applications G. Dethier 1, P. Marchot 2 and P.A. de Marneffe 1 1 EECS Department, University of Liege, Belgium 2 Chemical Engineering Department,

More information

Keywords: Dynamic Load Balancing, Process Migration, Load Indices, Threshold Level, Response Time, Process Age.

Keywords: Dynamic Load Balancing, Process Migration, Load Indices, Threshold Level, Response Time, Process Age. Volume 3, Issue 10, October 2013 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com Load Measurement

More information

LOAD BALANCING TECHNIQUES

LOAD BALANCING TECHNIQUES LOAD BALANCING TECHNIQUES Two imporatnt characteristics of distributed systems are resource multiplicity and system transparency. In a distributed system we have a number of resources interconnected by

More information

Experiments on the local load balancing algorithms; part 1

Experiments on the local load balancing algorithms; part 1 Experiments on the local load balancing algorithms; part 1 Ştefan Măruşter Institute e-austria Timisoara West University of Timişoara, Romania maruster@info.uvt.ro Abstract. In this paper the influence

More information

Scheduling Allowance Adaptability in Load Balancing technique for Distributed Systems

Scheduling Allowance Adaptability in Load Balancing technique for Distributed Systems Scheduling Allowance Adaptability in Load Balancing technique for Distributed Systems G.Rajina #1, P.Nagaraju #2 #1 M.Tech, Computer Science Engineering, TallaPadmavathi Engineering College, Warangal,

More information

DYNAMIC GRAPH ANALYSIS FOR LOAD BALANCING APPLICATIONS

DYNAMIC GRAPH ANALYSIS FOR LOAD BALANCING APPLICATIONS DYNAMIC GRAPH ANALYSIS FOR LOAD BALANCING APPLICATIONS DYNAMIC GRAPH ANALYSIS FOR LOAD BALANCING APPLICATIONS by Belal Ahmad Ibraheem Nwiran Dr. Ali Shatnawi Thesis submitted in partial fulfillment of

More information

How To Compare Load Sharing And Job Scheduling In A Network Of Workstations

How To Compare Load Sharing And Job Scheduling In A Network Of Workstations A COMPARISON OF LOAD SHARING AND JOB SCHEDULING IN A NETWORK OF WORKSTATIONS HELEN D. KARATZA Department of Informatics Aristotle University of Thessaloniki 546 Thessaloniki, GREECE Email: karatza@csd.auth.gr

More information

Load Balancing Algorithms for Peer to Peer and Client Server Distributed Environments

Load Balancing Algorithms for Peer to Peer and Client Server Distributed Environments Load Balancing Algorithms for Peer to Peer and Client Server Distributed Environments Sameena Naaz Afshar Alam Ranjit Biswas Department of Computer Science Jamia Hamdard, New Delhi, India ABSTRACT Advancements

More information

An Empirical Study and Analysis of the Dynamic Load Balancing Techniques Used in Parallel Computing Systems

An Empirical Study and Analysis of the Dynamic Load Balancing Techniques Used in Parallel Computing Systems An Empirical Study and Analysis of the Dynamic Load Balancing Techniques Used in Parallel Computing Systems Ardhendu Mandal and Subhas Chandra Pal Department of Computer Science and Application, University

More information

Distributed Explicit Partial Rerouting (DEPR) Scheme for Load Balancing in MPLS Networks

Distributed Explicit Partial Rerouting (DEPR) Scheme for Load Balancing in MPLS Networks Distributed Eplicit Partial Rerouting (DEPR) Scheme for Load Balancing in MPLS Networks Sherif Ibrahim Mohamed shf_ibrahim@yahoo.com Khaled M. F. Elsayed, senior member IEEE khaled@ieee.org Department

More information

Praktikum Wissenschaftliches Rechnen (Performance-optimized optimized Programming)

Praktikum Wissenschaftliches Rechnen (Performance-optimized optimized Programming) Praktikum Wissenschaftliches Rechnen (Performance-optimized optimized Programming) Dynamic Load Balancing Dr. Ralf-Peter Mundani Center for Simulation Technology in Engineering Technische Universität München

More information

A Novel Switch Mechanism for Load Balancing in Public Cloud

A Novel Switch Mechanism for Load Balancing in Public Cloud International OPEN ACCESS Journal Of Modern Engineering Research (IJMER) A Novel Switch Mechanism for Load Balancing in Public Cloud Kalathoti Rambabu 1, M. Chandra Sekhar 2 1 M. Tech (CSE), MVR College

More information

The Probabilistic Model of Cloud Computing

The Probabilistic Model of Cloud Computing A probabilistic multi-tenant model for virtual machine mapping in cloud systems Zhuoyao Wang, Majeed M. Hayat, Nasir Ghani, and Khaled B. Shaban Department of Electrical and Computer Engineering, University

More information

Key Words: Dynamic Load Balancing, and Distributed System

Key Words: Dynamic Load Balancing, and Distributed System DYNAMIC ROTATING LOAD BALANCING ALGORITHM IN DISTRIBUTED SYSTEMS ROSE SULEIMAN AL DAHOUD ALI ISSA OTOUM Al-Zaytoonah University Al-Zaytoonah University Neelain University rosesuleiman@yahoo.com aldahoud@alzaytoonah.edu.jo

More information

A Comparative Performance Analysis of Load Balancing Algorithms in Distributed System using Qualitative Parameters

A Comparative Performance Analysis of Load Balancing Algorithms in Distributed System using Qualitative Parameters A Comparative Performance Analysis of Load Balancing Algorithms in Distributed System using Qualitative Parameters Abhijit A. Rajguru, S.S. Apte Abstract - A distributed system can be viewed as a collection

More information

Dynamic load balancing of parallel cellular automata

Dynamic load balancing of parallel cellular automata Dynamic load balancing of parallel cellular automata Marc Mazzariol, Benoit A. Gennart, Roger D. Hersch Ecole Polytechnique Fédérale de Lausanne, EPFL * ABSTRACT We are interested in running in parallel

More information

IMPROVED PROXIMITY AWARE LOAD BALANCING FOR HETEROGENEOUS NODES

IMPROVED PROXIMITY AWARE LOAD BALANCING FOR HETEROGENEOUS NODES www.ijecs.in International Journal Of Engineering And Computer Science ISSN:2319-7242 Volume 2 Issue 6 June, 2013 Page No. 1914-1919 IMPROVED PROXIMITY AWARE LOAD BALANCING FOR HETEROGENEOUS NODES Ms.

More information

Fair Scheduling Algorithm with Dynamic Load Balancing Using In Grid Computing

Fair Scheduling Algorithm with Dynamic Load Balancing Using In Grid Computing Research Inventy: International Journal Of Engineering And Science Vol.2, Issue 10 (April 2013), Pp 53-57 Issn(e): 2278-4721, Issn(p):2319-6483, Www.Researchinventy.Com Fair Scheduling Algorithm with Dynamic

More information

Load Balancing. Load Balancing 1 / 24

Load Balancing. Load Balancing 1 / 24 Load Balancing Backtracking, branch & bound and alpha-beta pruning: how to assign work to idle processes without much communication? Additionally for alpha-beta pruning: implementing the young-brothers-wait

More information

Load Balancing Between Heterogenous Computing Clusters

Load Balancing Between Heterogenous Computing Clusters Load Balancing Between Heterogenous Computing Clusters Siu-Cheung Chau Dept. of Physics and Computing, Wilfrid Laurier University, Waterloo, Ontario, Canada, N2L 3C5 e-mail: schau@wlu.ca Ada Wai-Chee Fu

More information

Load balancing Static Load Balancing

Load balancing Static Load Balancing Chapter 7 Load Balancing and Termination Detection Load balancing used to distribute computations fairly across processors in order to obtain the highest possible execution speed. Termination detection

More information

An Ecient Dynamic Load Balancing using the Dimension Exchange. Ju-wook Jang. of balancing load among processors, most of the realworld

An Ecient Dynamic Load Balancing using the Dimension Exchange. Ju-wook Jang. of balancing load among processors, most of the realworld An Ecient Dynamic Load Balancing using the Dimension Exchange Method for Balancing of Quantized Loads on Hypercube Multiprocessors * Hwakyung Rim Dept. of Computer Science Seoul Korea 11-74 ackyung@arqlab1.sogang.ac.kr

More information

Performance of networks containing both MaxNet and SumNet links

Performance of networks containing both MaxNet and SumNet links Performance of networks containing both MaxNet and SumNet links Lachlan L. H. Andrew and Bartek P. Wydrowski Abstract Both MaxNet and SumNet are distributed congestion control architectures suitable for

More information

Dynamic Multi-User Load Balancing in Distributed Systems

Dynamic Multi-User Load Balancing in Distributed Systems Dynamic Multi-User Load Balancing in Distributed Systems Satish Penmatsa and Anthony T. Chronopoulos The University of Texas at San Antonio Dept. of Computer Science One UTSA Circle, San Antonio, Texas

More information

A Load Balancing Algorithm based on the Variation Trend of Entropy in Homogeneous Cluster

A Load Balancing Algorithm based on the Variation Trend of Entropy in Homogeneous Cluster , pp.11-20 http://dx.doi.org/10.14257/ ijgdc.2014.7.2.02 A Load Balancing Algorithm based on the Variation Trend of Entropy in Homogeneous Cluster Kehe Wu 1, Long Chen 2, Shichao Ye 2 and Yi Li 2 1 Beijing

More information

Dynamic Load Balancing in a Network of Workstations

Dynamic Load Balancing in a Network of Workstations Dynamic Load Balancing in a Network of Workstations 95.515F Research Report By: Shahzad Malik (219762) November 29, 2000 Table of Contents 1 Introduction 3 2 Load Balancing 4 2.1 Static Load Balancing

More information

RESEARCH PAPER International Journal of Recent Trends in Engineering, Vol 1, No. 1, May 2009

RESEARCH PAPER International Journal of Recent Trends in Engineering, Vol 1, No. 1, May 2009 An Algorithm for Dynamic Load Balancing in Distributed Systems with Multiple Supporting Nodes by Exploiting the Interrupt Service Parveen Jain 1, Daya Gupta 2 1,2 Delhi College of Engineering, New Delhi,

More information

Grid Computing Approach for Dynamic Load Balancing

Grid Computing Approach for Dynamic Load Balancing International Journal of Computer Sciences and Engineering Open Access Review Paper Volume-4, Issue-1 E-ISSN: 2347-2693 Grid Computing Approach for Dynamic Load Balancing Kapil B. Morey 1*, Sachin B. Jadhav

More information

Load Balancing and Termination Detection

Load Balancing and Termination Detection Chapter 7 Load Balancing and Termination Detection 1 Load balancing used to distribute computations fairly across processors in order to obtain the highest possible execution speed. Termination detection

More information

Dynamic Load Balancing in Charm++ Abhinav S Bhatele Parallel Programming Lab, UIUC

Dynamic Load Balancing in Charm++ Abhinav S Bhatele Parallel Programming Lab, UIUC Dynamic Load Balancing in Charm++ Abhinav S Bhatele Parallel Programming Lab, UIUC Outline Dynamic Load Balancing framework in Charm++ Measurement Based Load Balancing Examples: Hybrid Load Balancers Topology-aware

More information

An Overview of CORBA-Based Load Balancing

An Overview of CORBA-Based Load Balancing An Overview of CORBA-Based Load Balancing Jian Shu, Linlan Liu, Shaowen Song, Member, IEEE Department of Computer Science Nanchang Institute of Aero-Technology,Nanchang, Jiangxi, P.R.China 330034 dylan_cn@yahoo.com

More information

A Study on the Application of Existing Load Balancing Algorithms for Large, Dynamic, Heterogeneous Distributed Systems

A Study on the Application of Existing Load Balancing Algorithms for Large, Dynamic, Heterogeneous Distributed Systems A Study on the Application of Existing Load Balancing Algorithms for Large, Dynamic, Heterogeneous Distributed Systems RUPAM MUKHOPADHYAY, DIBYAJYOTI GHOSH AND NANDINI MUKHERJEE Department of Computer

More information

Various Schemes of Load Balancing in Distributed Systems- A Review

Various Schemes of Load Balancing in Distributed Systems- A Review 741 Various Schemes of Load Balancing in Distributed Systems- A Review Monika Kushwaha Pranveer Singh Institute of Technology Kanpur, U.P. (208020) U.P.T.U., Lucknow Saurabh Gupta Pranveer Singh Institute

More information

A Review of Customized Dynamic Load Balancing for a Network of Workstations

A Review of Customized Dynamic Load Balancing for a Network of Workstations A Review of Customized Dynamic Load Balancing for a Network of Workstations Taken from work done by: Mohammed Javeed Zaki, Wei Li, Srinivasan Parthasarathy Computer Science Department, University of Rochester

More information

Load Balancing and Termination Detection

Load Balancing and Termination Detection Chapter 7 slides7-1 Load Balancing and Termination Detection slides7-2 Load balancing used to distribute computations fairly across processors in order to obtain the highest possible execution speed. Termination

More information

Revenue Management for Transportation Problems

Revenue Management for Transportation Problems Revenue Management for Transportation Problems Francesca Guerriero Giovanna Miglionico Filomena Olivito Department of Electronic Informatics and Systems, University of Calabria Via P. Bucci, 87036 Rende

More information

A novel load balancing algorithm for computational grid

A novel load balancing algorithm for computational grid International Journal of Computational Intelligence Techniques, ISSN: 0976 0466 & E-ISSN: 0976 0474 Volume 1, Issue 1, 2010, PP-20-26 A novel load balancing algorithm for computational grid Saravanakumar

More information

Cellular Computing on a Linux Cluster

Cellular Computing on a Linux Cluster Cellular Computing on a Linux Cluster Alexei Agueev, Bernd Däne, Wolfgang Fengler TU Ilmenau, Department of Computer Architecture Topics 1. Cellular Computing 2. The Experiment 3. Experimental Results

More information

How To Balance In Cloud Computing

How To Balance In Cloud Computing A Review on Load Balancing Algorithms in Cloud Hareesh M J Dept. of CSE, RSET, Kochi hareeshmjoseph@ gmail.com John P Martin Dept. of CSE, RSET, Kochi johnpm12@gmail.com Yedhu Sastri Dept. of IT, RSET,

More information

Performance Comparison of Dynamic Load-Balancing Strategies for Distributed Computing

Performance Comparison of Dynamic Load-Balancing Strategies for Distributed Computing Performance Comparison of Dynamic Load-Balancing Strategies for Distributed Computing A. Cortés, A. Ripoll, M.A. Senar and E. Luque Computer Architecture and Operating Systems Group Universitat Autònoma

More information

International journal of Engineering Research-Online A Peer Reviewed International Journal Articles available online http://www.ijoer.

International journal of Engineering Research-Online A Peer Reviewed International Journal Articles available online http://www.ijoer. RESEARCH ARTICLE ISSN: 2321-7758 GLOBAL LOAD DISTRIBUTION USING SKIP GRAPH, BATON AND CHORD J.K.JEEVITHA, B.KARTHIKA* Information Technology,PSNA College of Engineering & Technology, Dindigul, India Article

More information

Load balancing; Termination detection

Load balancing; Termination detection Load balancing; Termination detection Parallel and Distributed Computing Department of Computer Science and Engineering (DEI) Instituto Superior Técnico November 14, 2013 CPD (DEI / IST) Parallel and Distributed

More information

How To Develop A Dynamic Load Balancing Algorithm

How To Develop A Dynamic Load Balancing Algorithm IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.6, June 2010 153 A Guide to Dynamic Load Balancing in Distributed Computer Systems Ali M. Alakeel College of Computing and

More information

A Simultaneous Solution for General Linear Equations on a Ring or Hierarchical Cluster

A Simultaneous Solution for General Linear Equations on a Ring or Hierarchical Cluster Acta Technica Jaurinensis Vol. 3. No. 1. 010 A Simultaneous Solution for General Linear Equations on a Ring or Hierarchical Cluster G. Molnárka, N. Varjasi Széchenyi István University Győr, Hungary, H-906

More information

A Robust Dynamic Load-balancing Scheme for Data Parallel Application on Message Passing Architecture

A Robust Dynamic Load-balancing Scheme for Data Parallel Application on Message Passing Architecture A Robust Dynamic Load-balancing Scheme for Data Parallel Application on Message Passing Architecture Yangsuk Kee Department of Computer Engineering Seoul National University Seoul, 151-742, Korea Soonhoi

More information

Load Distribution in Large Scale Network Monitoring Infrastructures

Load Distribution in Large Scale Network Monitoring Infrastructures Load Distribution in Large Scale Network Monitoring Infrastructures Josep Sanjuàs-Cuxart, Pere Barlet-Ros, Gianluca Iannaccone, and Josep Solé-Pareta Universitat Politècnica de Catalunya (UPC) {jsanjuas,pbarlet,pareta}@ac.upc.edu

More information

DYNAMIC LOAD BALANCING IN A DECENTRALISED DISTRIBUTED SYSTEM

DYNAMIC LOAD BALANCING IN A DECENTRALISED DISTRIBUTED SYSTEM DYNAMIC LOAD BALANCING IN A DECENTRALISED DISTRIBUTED SYSTEM 1 Introduction In parallel distributed computing system, due to the lightly loaded and overloaded nodes that cause load imbalance, could affect

More information

CHAPTER 5 WLDMA: A NEW LOAD BALANCING STRATEGY FOR WAN ENVIRONMENT

CHAPTER 5 WLDMA: A NEW LOAD BALANCING STRATEGY FOR WAN ENVIRONMENT 81 CHAPTER 5 WLDMA: A NEW LOAD BALANCING STRATEGY FOR WAN ENVIRONMENT 5.1 INTRODUCTION Distributed Web servers on the Internet require high scalability and availability to provide efficient services to

More information

Design of an Optimized Virtual Server for Efficient Management of Cloud Load in Multiple Cloud Environments

Design of an Optimized Virtual Server for Efficient Management of Cloud Load in Multiple Cloud Environments Design of an Optimized Virtual Server for Efficient Management of Cloud Load in Multiple Cloud Environments Ajay A. Jaiswal 1, Dr. S. K. Shriwastava 2 1 Associate Professor, Department of Computer Technology

More information

Static Load Balancing of Parallel PDE Solver for Distributed Computing Environment

Static Load Balancing of Parallel PDE Solver for Distributed Computing Environment Static Load Balancing of Parallel PDE Solver for Distributed Computing Environment Shuichi Ichikawa and Shinji Yamashita Department of Knowledge-based Information Engineering, Toyohashi University of Technology

More information

A New Nature-inspired Algorithm for Load Balancing

A New Nature-inspired Algorithm for Load Balancing A New Nature-inspired Algorithm for Load Balancing Xiang Feng East China University of Science and Technology Shanghai, China 200237 Email: xfeng{@ecusteducn, @cshkuhk} Francis CM Lau The University of

More information

Network Security Validation Using Game Theory

Network Security Validation Using Game Theory Network Security Validation Using Game Theory Vicky Papadopoulou and Andreas Gregoriades Computer Science and Engineering Dep., European University Cyprus, Cyprus {v.papadopoulou,a.gregoriades}@euc.ac.cy

More information

Self-organized Multi-agent System for Service Management in the Next Generation Networks

Self-organized Multi-agent System for Service Management in the Next Generation Networks PROCEEDINGS OF THE WORKSHOP ON APPLICATIONS OF SOFTWARE AGENTS ISBN 978-86-7031-188-6, pp. 18-24, 2011 Self-organized Multi-agent System for Service Management in the Next Generation Networks Mario Kusek

More information

SOLVING LOAD REBALANCING FOR DISTRIBUTED FILE SYSTEM IN CLOUD

SOLVING LOAD REBALANCING FOR DISTRIBUTED FILE SYSTEM IN CLOUD International Journal of Advances in Applied Science and Engineering (IJAEAS) ISSN (P): 2348-1811; ISSN (E): 2348-182X Vol-1, Iss.-3, JUNE 2014, 54-58 IIST SOLVING LOAD REBALANCING FOR DISTRIBUTED FILE

More information

Asynchronous Computations

Asynchronous Computations Asynchronous Computations Asynchronous Computations Computations in which individual processes operate without needing to synchronize with other processes. Synchronizing processes is an expensive operation

More information

Load Balancing Techniques

Load Balancing Techniques Load Balancing Techniques 1 Lecture Outline Following Topics will be discussed Static Load Balancing Dynamic Load Balancing Mapping for load balancing Minimizing Interaction 2 1 Load Balancing Techniques

More information

Preserving Message Integrity in Dynamic Process Migration

Preserving Message Integrity in Dynamic Process Migration Preserving Message Integrity in Dynamic Process Migration E. Heymann, F. Tinetti, E. Luque Universidad Autónoma de Barcelona Departamento de Informática 8193 - Bellaterra, Barcelona, Spain e-mail: e.heymann@cc.uab.es

More information

Index Terms : Load rebalance, distributed file systems, clouds, movement cost, load imbalance, chunk.

Index Terms : Load rebalance, distributed file systems, clouds, movement cost, load imbalance, chunk. Load Rebalancing for Distributed File Systems in Clouds. Smita Salunkhe, S. S. Sannakki Department of Computer Science and Engineering KLS Gogte Institute of Technology, Belgaum, Karnataka, India Affiliated

More information

How To Understand The Concept Of A Distributed System

How To Understand The Concept Of A Distributed System Distributed Operating Systems Introduction Ewa Niewiadomska-Szynkiewicz and Adam Kozakiewicz ens@ia.pw.edu.pl, akozakie@ia.pw.edu.pl Institute of Control and Computation Engineering Warsaw University of

More information

Load Balancing and Termination Detection

Load Balancing and Termination Detection Chapter 7 Slide 1 Slide 2 Load Balancing and Termination Detection Load balancing used to distribute computations fairly across processors in order to obtain the highest possible execution speed. Termination

More information

Load Balancing between Computing Clusters

Load Balancing between Computing Clusters Load Balancing between Computing Clusters Siu-Cheung Chau Dept. of Physics and Computing, Wilfrid Laurier University, Waterloo, Ontario, Canada, NL 3C5 e-mail: schau@wlu.ca Ada Wai-Chee Fu Dept. of Computer

More information

Path Selection Methods for Localized Quality of Service Routing

Path Selection Methods for Localized Quality of Service Routing Path Selection Methods for Localized Quality of Service Routing Xin Yuan and Arif Saifee Department of Computer Science, Florida State University, Tallahassee, FL Abstract Localized Quality of Service

More information

Chapter 7 Load Balancing and Termination Detection

Chapter 7 Load Balancing and Termination Detection Chapter 7 Load Balancing and Termination Detection Load balancing used to distribute computations fairly across processors in order to obtain the highest possible execution speed. Termination detection

More information

A Tool for Generating Partition Schedules of Multiprocessor Systems

A Tool for Generating Partition Schedules of Multiprocessor Systems A Tool for Generating Partition Schedules of Multiprocessor Systems Hans-Joachim Goltz and Norbert Pieth Fraunhofer FIRST, Berlin, Germany {hans-joachim.goltz,nobert.pieth}@first.fraunhofer.de Abstract.

More information

How To Balance In A Distributed System

How To Balance In A Distributed System 6 IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 11, NO. 1, JANUARY 2000 How Useful Is Old Information? Michael Mitzenmacher AbstractÐWe consider the problem of load balancing in dynamic distributed

More information

DYNAMIC LOAD BALANCING SCHEME FOR ITERATIVE APPLICATIONS

DYNAMIC LOAD BALANCING SCHEME FOR ITERATIVE APPLICATIONS Journal homepage: www.mjret.in DYNAMIC LOAD BALANCING SCHEME FOR ITERATIVE APPLICATIONS ISSN:2348-6953 Rahul S. Wankhade, Darshan M. Marathe, Girish P. Nikam, Milind R. Jawale Department of Computer Engineering,

More information

GEDAE TM - A Graphical Programming and Autocode Generation Tool for Signal Processor Applications

GEDAE TM - A Graphical Programming and Autocode Generation Tool for Signal Processor Applications GEDAE TM - A Graphical Programming and Autocode Generation Tool for Signal Processor Applications Harris Z. Zebrowitz Lockheed Martin Advanced Technology Laboratories 1 Federal Street Camden, NJ 08102

More information

An Interactive Visualization Tool for the Analysis of Multi-Objective Embedded Systems Design Space Exploration

An Interactive Visualization Tool for the Analysis of Multi-Objective Embedded Systems Design Space Exploration An Interactive Visualization Tool for the Analysis of Multi-Objective Embedded Systems Design Space Exploration Toktam Taghavi, Andy D. Pimentel Computer Systems Architecture Group, Informatics Institute

More information

LOAD BALANCING WITH PARTIAL KNOWLEDGE OF SYSTEM

LOAD BALANCING WITH PARTIAL KNOWLEDGE OF SYSTEM LOAD BALANCING WITH PARTIAL KNOWLEDGE OF SYSTEM IN PEER TO PEER NETWORKS R. Vijayalakshmi and S. Muthu Kumarasamy Dept. of Computer Science & Engineering, S.A. Engineering College Anna University, Chennai,

More information

Introduction to LAN/WAN. Network Layer

Introduction to LAN/WAN. Network Layer Introduction to LAN/WAN Network Layer Topics Introduction (5-5.1) Routing (5.2) (The core) Internetworking (5.5) Congestion Control (5.3) Network Layer Design Isues Store-and-Forward Packet Switching Services

More information

Keywords Load balancing, Dispatcher, Distributed Cluster Server, Static Load balancing, Dynamic Load balancing.

Keywords Load balancing, Dispatcher, Distributed Cluster Server, Static Load balancing, Dynamic Load balancing. Volume 5, Issue 7, July 2015 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com A Hybrid Algorithm

More information

Principles and characteristics of distributed systems and environments

Principles and characteristics of distributed systems and environments Principles and characteristics of distributed systems and environments Definition of a distributed system Distributed system is a collection of independent computers that appears to its users as a single

More information

Dynamic Resource Allocation in Softwaredefined Radio The Interrelation Between Platform Architecture and Application Mapping

Dynamic Resource Allocation in Softwaredefined Radio The Interrelation Between Platform Architecture and Application Mapping Dynamic Resource Allocation in Softwaredefined Radio The Interrelation Between Platform Architecture and Application Mapping V. Marojevic, X. Revés, A. Gelonch Polythechnic University of Catalonia Dept.

More information

A Fast Path Recovery Mechanism for MPLS Networks

A Fast Path Recovery Mechanism for MPLS Networks A Fast Path Recovery Mechanism for MPLS Networks Jenhui Chen, Chung-Ching Chiou, and Shih-Lin Wu Department of Computer Science and Information Engineering Chang Gung University, Taoyuan, Taiwan, R.O.C.

More information

Characterizing the Performance of Dynamic Distribution and Load-Balancing Techniques for Adaptive Grid Hierarchies

Characterizing the Performance of Dynamic Distribution and Load-Balancing Techniques for Adaptive Grid Hierarchies Proceedings of the IASTED International Conference Parallel and Distributed Computing and Systems November 3-6, 1999 in Cambridge Massachusetts, USA Characterizing the Performance of Dynamic Distribution

More information

Load Balancing on a Non-dedicated Heterogeneous Network of Workstations

Load Balancing on a Non-dedicated Heterogeneous Network of Workstations Load Balancing on a Non-dedicated Heterogeneous Network of Workstations Dr. Maurice Eggen Nathan Franklin Department of Computer Science Trinity University San Antonio, Texas 78212 Dr. Roger Eggen Department

More information

A Performance Study of Load Balancing Strategies for Approximate String Matching on an MPI Heterogeneous System Environment

A Performance Study of Load Balancing Strategies for Approximate String Matching on an MPI Heterogeneous System Environment A Performance Study of Load Balancing Strategies for Approximate String Matching on an MPI Heterogeneous System Environment Panagiotis D. Michailidis and Konstantinos G. Margaritis Parallel and Distributed

More information

PERFORMANCE EVALUATION OF THREE DYNAMIC LOAD BALANCING ALGORITHMS ON SPMD MODEL

PERFORMANCE EVALUATION OF THREE DYNAMIC LOAD BALANCING ALGORITHMS ON SPMD MODEL PERFORMANCE EVALUATION OF THREE DYNAMIC LOAD BALANCING ALGORITHMS ON SPMD MODEL Najib A. Kofahi Associate Professor Department of Computer Sciences Faculty of Information Technology and Computer Sciences

More information

In Proceedings of the First IEEE Workshop on Real-Time Applications, New York, NY, May 1993.

In Proceedings of the First IEEE Workshop on Real-Time Applications, New York, NY, May 1993. In Proceedings of the First IEEE Workshop on Real-Time Applications, New York, NY, May 1993. Probabilistic Job Scheduling for Distributed Real-time Applications Azer Bestavros Dimitrios Spartiotis Computer

More information

A SIMULATOR FOR LOAD BALANCING ANALYSIS IN DISTRIBUTED SYSTEMS

A SIMULATOR FOR LOAD BALANCING ANALYSIS IN DISTRIBUTED SYSTEMS Mihai Horia Zaharia, Florin Leon, Dan Galea (3) A Simulator for Load Balancing Analysis in Distributed Systems in A. Valachi, D. Galea, A. M. Florea, M. Craus (eds.) - Tehnologii informationale, Editura

More information

On-line scheduling algorithm for real-time multiprocessor systems with ACO

On-line scheduling algorithm for real-time multiprocessor systems with ACO International Journal of Intelligent Information Systems 2015; 4(2-1): 13-17 Published online January 28, 2015 (http://www.sciencepublishinggroup.com/j/ijiis) doi: 10.11648/j.ijiis.s.2015040201.13 ISSN:

More information

Adaptive Load Balancing Method Enabling Auto-Specifying Threshold of Node Load Status for Apache Flume

Adaptive Load Balancing Method Enabling Auto-Specifying Threshold of Node Load Status for Apache Flume , pp. 201-210 http://dx.doi.org/10.14257/ijseia.2015.9.2.17 Adaptive Load Balancing Method Enabling Auto-Specifying Threshold of Node Load Status for Apache Flume UnGyu Han and Jinho Ahn Dept. of Comp.

More information

CSE 4351/5351 Notes 7: Task Scheduling & Load Balancing

CSE 4351/5351 Notes 7: Task Scheduling & Load Balancing CSE / Notes : Task Scheduling & Load Balancing Task Scheduling A task is a (sequential) activity that uses a set of inputs to produce a set of outputs. A task (precedence) graph is an acyclic, directed

More information

A Locality Enhanced Scheduling Method for Multiple MapReduce Jobs In a Workflow Application

A Locality Enhanced Scheduling Method for Multiple MapReduce Jobs In a Workflow Application 2012 International Conference on Information and Computer Applications (ICICA 2012) IPCSIT vol. 24 (2012) (2012) IACSIT Press, Singapore A Locality Enhanced Scheduling Method for Multiple MapReduce Jobs

More information

Parallel Job Scheduling in Homogeneous Distributed Systems

Parallel Job Scheduling in Homogeneous Distributed Systems Parallel Job Scheduling in Homogeneous Distributed Systems Helen D. Karatza Department of Informatics Aristotle University of Thessaloniki 51 Thessaloniki, Greece Ralph C. Hilzer Computer Science Department

More information

Sla Aware Load Balancing Algorithm Using Join-Idle Queue for Virtual Machines in Cloud Computing

Sla Aware Load Balancing Algorithm Using Join-Idle Queue for Virtual Machines in Cloud Computing Sla Aware Load Balancing Using Join-Idle Queue for Virtual Machines in Cloud Computing Mehak Choudhary M.Tech Student [CSE], Dept. of CSE, SKIET, Kurukshetra University, Haryana, India ABSTRACT: Cloud

More information

Operatin g Systems: Internals and Design Principle s. Chapter 10 Multiprocessor and Real-Time Scheduling Seventh Edition By William Stallings

Operatin g Systems: Internals and Design Principle s. Chapter 10 Multiprocessor and Real-Time Scheduling Seventh Edition By William Stallings Operatin g Systems: Internals and Design Principle s Chapter 10 Multiprocessor and Real-Time Scheduling Seventh Edition By William Stallings Operating Systems: Internals and Design Principles Bear in mind,

More information

Dynamic Load Balance Algorithm (DLBA) for IEEE 802.11 Wireless LAN

Dynamic Load Balance Algorithm (DLBA) for IEEE 802.11 Wireless LAN Tamkang Journal of Science and Engineering, vol. 2, No. 1 pp. 45-52 (1999) 45 Dynamic Load Balance Algorithm () for IEEE 802.11 Wireless LAN Shiann-Tsong Sheu and Chih-Chiang Wu Department of Electrical

More information

Load Balancing to Save Energy in Cloud Computing

Load Balancing to Save Energy in Cloud Computing presented at the Energy Efficient Systems Workshop at ICT4S, Stockholm, Aug. 2014 Load Balancing to Save Energy in Cloud Computing Theodore Pertsas University of Manchester United Kingdom tpertsas@gmail.com

More information

A Service Revenue-oriented Task Scheduling Model of Cloud Computing

A Service Revenue-oriented Task Scheduling Model of Cloud Computing Journal of Information & Computational Science 10:10 (2013) 3153 3161 July 1, 2013 Available at http://www.joics.com A Service Revenue-oriented Task Scheduling Model of Cloud Computing Jianguang Deng a,b,,

More information

Resource Allocation Schemes for Gang Scheduling

Resource Allocation Schemes for Gang Scheduling Resource Allocation Schemes for Gang Scheduling B. B. Zhou School of Computing and Mathematics Deakin University Geelong, VIC 327, Australia D. Walsh R. P. Brent Department of Computer Science Australian

More information

Lecture Outline Overview of real-time scheduling algorithms Outline relative strengths, weaknesses

Lecture Outline Overview of real-time scheduling algorithms Outline relative strengths, weaknesses Overview of Real-Time Scheduling Embedded Real-Time Software Lecture 3 Lecture Outline Overview of real-time scheduling algorithms Clock-driven Weighted round-robin Priority-driven Dynamic vs. static Deadline

More information

Figure 1. The cloud scales: Amazon EC2 growth [2].

Figure 1. The cloud scales: Amazon EC2 growth [2]. - Chung-Cheng Li and Kuochen Wang Department of Computer Science National Chiao Tung University Hsinchu, Taiwan 300 shinji10343@hotmail.com, kwang@cs.nctu.edu.tw Abstract One of the most important issues

More information

A Content-Based Load Balancing Algorithm for Metadata Servers in Cluster File Systems*

A Content-Based Load Balancing Algorithm for Metadata Servers in Cluster File Systems* A Content-Based Load Balancing Algorithm for Metadata Servers in Cluster File Systems* Junho Jang, Saeyoung Han, Sungyong Park, and Jihoon Yang Department of Computer Science and Interdisciplinary Program

More information

PARALLEL PROGRAMMING

PARALLEL PROGRAMMING PARALLEL PROGRAMMING TECHNIQUES AND APPLICATIONS USING NETWORKED WORKSTATIONS AND PARALLEL COMPUTERS 2nd Edition BARRY WILKINSON University of North Carolina at Charlotte Western Carolina University MICHAEL

More information

Comparison on Different Load Balancing Algorithms of Peer to Peer Networks

Comparison on Different Load Balancing Algorithms of Peer to Peer Networks Comparison on Different Load Balancing Algorithms of Peer to Peer Networks K.N.Sirisha *, S.Bhagya Rekha M.Tech,Software Engineering Noble college of Engineering & Technology for Women Web Technologies

More information

Load balancing; Termination detection

Load balancing; Termination detection Load balancing; Termination detection Parallel and Distributed Computing Department of Computer Science and Engineering (DEI) Instituto Superior Técnico November 13, 2014 CPD (DEI / IST) Parallel and Distributed

More information