Handling Single Node Failures Using Agents in Computer Clusters

Size: px
Start display at page:

Download "Handling Single Node Failures Using Agents in Computer Clusters"

Transcription

1 Handling Single Node Failures Using Agents in Computer Clusters Blesson Varghese, Gerard McKee and Vassil Alexandrov School of Systems Engineering, University of Reading Whiteknights Campus, Reading, Berkshire United Kingdom, RG6 6AY Abstract The work reported in this paper is motivated towards handling single node failures for parallel summation algorithms in computer clusters. An agent based approach is proposed in which a task to be executed is decomposed to subtasks and mapped onto agents that traverse computing nodes. The agents intercommunicate across computing nodes to share information during the event of a predicted node failure. Two single node failure scenarios are considered. The Message Passing Interface is employed for implementing the proposed approach. Quantitative results obtained from experiments reveal that the agent based approach can handle failures more efficiently than traditional failure handling approaches. Index Terms Failure handling; Cluster computing; Message Passing Interface; Agent-based failure handling; Single node failure I. INTRODUCTION Proactive failure handling forms a crucial component of research in fault tolerance for distributed parallel computing systems. Handling failures proactively, as the term implies is the prediction of failures on computing nodes and moves tasks from such computing nodes predicted to fail onto safe computing nodes less likely to fail soon [1][2][3][4]. Hence, proactively handling failures aim for controlling a situation by causing something to happen rather than waiting to respond after it happens. Traditional approaches to handle failures include methods such as checkpointing, replication and message logging, and are reported in research that focuses on fault tolerance of distributed parallel computing systems. However, the traditional failure handling approaches are challenged by single point failures, scalability issues, communication overheads and prolonged periods of time for reinstating process execution [5][6]. In more recent times, agent technology has been employed for failure handling. Approaches that employ agent technology have incorporated failure handling strategies that tend to be more dynamic and address many issues that are a challenge for traditional failure handling approaches. Research based on agent based failure handling can classified as failure handling by an agent framework and failure handling by individual agents within an agent framework. Both researches have significantly contributed to achieve agent based failure handling. Failure handling by agent frameworks is reported in [7][8] and [9]. Tichy et al. [7] identify failure handling multiagent system characteristics and consider a potential framework, namely Autonomous Cooperative System (ACS). Key concepts of the framework include reliable communication, fault-tolerant agent platform, fault-tolerant social knowledge, physical distribution and fault-tolerant agent architecture. Mendes et al. [8] propose a fault tolerant networked control system. In the proposed system, the number of critical communications needed for safe operations between system components are minimized, hence guaranteeing safe operation and performance even in faulty conditions. Almeida et al. [9] report the implementation of the Dynamic Agent Replication Extension (DARX), a failure handling agent framework. In this model, failure handling is performed by replicating those agents that are critical to the system and whose future plans could influence other agents in the system. Failure handling by individual agents within an agent framework is reported in [10] and [11]. Khan et al. [10] propose exception handling and periodic events that are sent to agents to inspect their state, though an overhead to the system, as a means to handle failures. The agent and broker system is integrated into the architecture to achieve failure handling capabilities in the brokerage system. Summiya et al. [11] address the prevention of partial or complete loss of an agent in an agent framework by employing an algorithm similar to the sliding window model [12]. A selected set of research based on failure handling using agent technology whose review is reported above, considers failure handling by an agent framework and agents within an agent framework. However, such research seldom explores the extension and implementation of such ideas for large scale distributed parallel computing systems. Hence, there exists a need to address the issue of failure handling employing agent technology in parallel computing systems. Failures of computing nodes in parallel computing systems can be classified as single node and multiple node failures. Single node failures occur when one node in an array of computing nodes have failed, while multiple node failures occur when more than one node in an array of computing nodes fail. The work reported in this paper considers only single node failures by prediction, on similar lines to proactive failure

2 handling. An agent based approach is proposed in which a task to be executed is decomposed to sub-tasks and mapped onto agents that traverse an abstracted hardware layer. The agents intercommunicate across processors to share information during the event of a predicted node failure and for successfully completing the task. The agents hence contribute towards handling faults efficiently. The remainder of this paper is organized as follows. Section 2 considers failure handling agents and their cognitive capabilities. Section 3 presents the implementation of the agent based approach for failure handling. The resources required for implementing the failure handling approach and how these resources are glued together coherently to handle failures is presented. Two scenarios of single node failure are also considered. Section 4 presents quantitative results based on experiments performed for the single node failure scenarios. Section 5 concludes the paper by considering future work. II. FAILURE HANDLING AGENTS Agent-based approaches are biologically inspired from nature. For example, swarming of agents in multi-agent systems like swarm robotic systems are inspired from the biological phenomena of swarming bees. Agents in a natural swarm also demonstrate intelligence by their cognitive capabilities in atleast four different ways [13][14]. Firstly, an agent is capable of being able to know its environment, the surroundings in which it is located. Secondly, an agent is capable to identify a location in the environment in which it can nicely situate. Thirdly, an agent is capable to sense any hazard that is likely to deteriorate or impair its functioning. Fourthly, an agent is capable to traverse from one location to another when necessary for survival. These four capabilities are also desirable for agents in a computing environment. The aim of the agent based approach for failure handing considered in this section is to achieve agent intelligence in parallel computing systems and further demonstrate that the cognitive capabilities of an agent complementing its intelligence can lead towards effective failure handling. In abstract terms the agent based approach proposed in this paper can be summarised as follows. A task to be executed on a parallel computing system is decomposed into subtasks and mapped onto agents that carry these tasks onto computing nodes for execution. The agent and the sub-problem are independent of each other; in other words, the agents only carry the sub-tasks or act as a wrapper around the sub-task independent of the operations performed by the task. The agents displace through the nodes to find an appropriate area to cluster and execute the task. In the proposed approach, an agent possesses capabilities similar to the capabilities of a natural agent presented above. Intelligence of an agent in the computing environment is demonstrated in four different ways. Firstly, an agent is aware of its environment that is the computing nodes on which it can carry a task onto, other agents in its vicinity and agents with which it interacts or shares information. Secondly, an agent can situate itself on a node that may not fail soon and can provide necessary and sufficient consistency in executing the task. Thirdly, an agent can predict node failures by consistent monitoring (for example, power consumption and heat dissipation of the nodes can be used to predict failures). Fourthly, an agent is capable of shifting gracefully from one node to another without causing interruption to the state of execution and notifying other interacting agents in the system when a node on which a sub-task being executed is predicted to fail. III. IMPLEMENTATION To implement the agent based approach considered in section 2 there is a requirement to consider the resources and how these resources can be glued together. Four resources, namely the executed problem, the parallel computing platform, the middleware and the hardware abstraction are considered. A. Resources Firstly, the executed problem an important aspect in large scale parallel computing systems is considered [15]. Parallel reduction algorithms are identified as a class of algorithms that can benefit from the agent based failure handling approach due to two reasons. Firstly, the computing nodes of a parallel reduction algorithm tend to be critical. The execution of the algorithm stalls or produces an incorrect solution if any node information is lost. Secondly, parallel reduction algorithms are employed in critical applications such as space applications. These applications need failure handling through self-managing real time systems. Parallel summation is an exemplar of parallel reduction algorithm and is considered as the executed problem in this paper. Figure 1 is an illustration of the parallel summation algorithm. The problem of addition is sub-divided between nodes as shown in the diagram, thereby generating sub-problems. These sub-problems are executed on parallel nodes for a given level, but executed sequentially on nodes between different levels. Secondly, the parallel computing platform to execute the problem is considered. In the research reported in this paper, a computer cluster is chosen as a platform for implementing the agent based approach for two reasons. Firstly, a cluster is often characterized by three basic elements, namely a collection of nodes, a network connecting these nodes and a facility to access and share information between the nodes [16], which are simpler elements to handle when compared to other parallel computing infrastructures. Secondly, existing middleware for clusters, namely Message Passing Interface (MPI)[17] provide standard and portable programming interfaces. The cluster used for the research reported in this paper is one among the high performance computing resources available at the Centre for Advanced Computing and Emerging Technologies (ACET), University of Reading, United Kingdom [18][19]. The cluster consists of a head node and 33 compute nodes. All nodes are connected via a Gigabit ethernet switch and communicate via the standard TCP protocol.

3 Fig. 2. Mapping hardware nodes to logical nodes Fig. 1. Illustration of the Parallel Summation Algorithm Thirdly, the middleware, for which Message Passing Interface, a standardized application programming interface (API) used for parallel and/or distributed computing, is chosen for implementing the agent based failure handling approach. Open MPI [20] [21] version 1.3.3, an open source implementation of MPI 2.0 is employed on the cluster. An important feature of MPI 2.0, dynamic process creation and management is essential for implementing the approach. The MPI dynamic process model permits the creation and management of a set of processes both when an MPI application begins and after the application has started. The management of newly created processes includes cooperative termination of a process, communication between newly created processes and existing MPI application, and establishing communication between two independent processes. MPI COMM SPAWN is used to create a new MPI process and establish communication from an existing MPI application. On the other hand, MPI COMM ACCEPT and MPI COMM CONNECT can be used to establish communication between two independent processes. More MPI specific details on dynamic process model can be obtained from [17] [22]. Fourthly, the hardware abstraction which is obtained when hardware nodes are abstracted to logical nodes. The hardware layer comprises physical nodes of the cluster that are connected via a switch, thereby forming a fully connected mesh topology. The abstracted layer is obtained when the physical nodes are abstracted to logical nodes, and is achieved by implementing software rules/policies using the middleware. The policies are such that a carrier agent carrying an executing sub-task can only communicate with a vertically, horizontally or diagonally adjacent carrier agent or process, effectively leading to a grid topology on the abstracted layer. For example, nine nodes of a computer cluster forming a fully connected mesh topology in figure 2 is abstracted to a grid topology in the abstraction layer. B. Gluing the Resources Having considered the resources there is a need to glue them together in a coherent fashion to achieve the goals of the agent based approach. This section hence considers how the resources considered in the previous sub-section are glued together. The parallel summation algorithm works in four sequential levels. The first level comprising nodes N 1 N 8 receives a live input feed of data. The second level comprising nodes N 9 N 12 receives data from the first level, adds the data received and yields the result to the third level nodes N 13 and N 14. The fourth level, adds data received from the third level nodes and produces the final result. Figure 1 shows the nodes required in the parallel summation algorithm. For a given time step, every node in a level operates in parallel. Each node is characterized by input dependencies (process or processor a node is dependent on for receiving an input), output dependencies (process or processor a node yields data to as output) and data contained in the node. The first level nodes have one input dependency and one output dependency. For instance, node N 1 has one input dependency I 1 and node N 9 as its output dependency. However, the second, third and fourth levels have two input dependencies and one output dependency. For instance, node N 13 of the third level has nodes N 9 and N 10 as input dependencies and node N 15 as output dependency. The data contained in a node is either the input data for the first level nodes or a calculated value (sum of two value in the case of a parallel summation algorithm) stored within a node. The agents on the abstracted layer are created such that they carry input and output dependencies and data. Since, parallel summation is relatively less complex when compared to other computational algorithms; the agents carry little information and have few dependencies. Agent intelligence is demonstrated in four different ways. Firstly, an agent is capable of being able to know its environment, the surroundings in which it is located. Information

4 concerning the environment includes knowledge about the processing node on which it is situated, knowledge about other processing nodes in its vicinity and knowledge about other agents situated on processing nodes located to its vicinity. Secondly, an agent is capable to identify a location in the environment in which it can nicely situate. As an agent continues to gather and update information about its vicinity, the agent is also capable to decide onto which processing node it can situate when the processing node on which it is currently situated is likely to fail. Thirdly, an agent is capable to sense any hazard that is likely to deteriorate or impair its functioning. This capability is on similar lines of proactive fault tolerance whereby a failure is predicted. In the intelligent agent based approach presented in this dissertation, rising temperatures of a processing node beyond a threshold is the factor that can impair the functioning of an agent. Hence, an agent is capable to sense this hazard Fourthly, an agent is capable to pass over from one location to another when necessary for survival. If a hazard is sensed the agent can relocate on another processing node and complete the execution of the task it is carrying. These four capabilities are utilized in the intelligent agent based approach. The agent capabilities are combined together to achieve the failure handling in the following manner. Each process executing on a node gathers some sensory information to predict whether a node is likely to fail, on similar lines to proactive fault tolerance. In the implementation presented in this paper node temperatures are simulated. When the temperature of a node rises beyond a threshold, the process executing on that node predicts a failure and hence spawns a process on an adjacent node in the abstracted layer. The agent on the abstracted node expected to fail shifts to the adjacent node on which the new process was spawned. The dependency information carried by the agent that was shifted to the new node is employed to reinstate the state of execution of the algorithm. The data for summation contained in the agent, either obtained from a previous level or a calculated value to be yielded to the next level, ensures that information is not lost and does not affect the final solution in critical applications. Though a preliminary implementation model was achieved, it was observed that MPI was not the most appropriate middleware for implementing the multi-agent approach. When an agent predicted a node failure, a new process had to be dynamically created on an adjacent node that was not predicted to fail, hence allowing the agent on the node predicted to fail to transfer control onto the agent on the newly created process. For this, MPI Comm spawn, MPI Comm connect and MPI Comm accept were required. Since some of these functionalities provided unstable results on the cluster used for implementation, a work-around had to be sought. Hence the process on the new node onto which the agent transferred was created during the initialization of the program and ran on the cluster as a dummy process until it came to play. Two scenarios based on single node failure (only one node fails in an instant) are considered in the implementation of the agent based approach. Fig. 3. Communication sequence for single node failure scenario 1 Fig. 4. Communication sequence for single node failure scenario 2 1) Single Node Failure Scenario 1: Firstly, a scenario in which no nodes connected to a node predicted to fail would fail in a consecutive time step. Figure 3 illustrates the sequence of events and communication in the first scenario. The communication sequences for the first scenario are as follows. Firstly, the hardware probing process of the node predicted to fail notifies the carrier agent situated on that node that it has predicted a failure. The carrier agent immediately spawns a new process on a node adjacent to it. Further to this the carrier agent sends notification to the input dependent processes (two processes in the case of parallel summation algorithm considered in this paper) and the output dependent process (one process in the case of parallel summation algorithm considered in this paper). After sending the notification to the dependent processes the agent process terminates execution. The newly spawned process then reestablishes all input and output dependencies and continues execution.

5 2) Single Node Failure Scenario 2: Secondly, a scenario that was more realistic in nature and assumed that any node connected to a node predicted to fail could also fail in a consecutive time step. Figure 4 illustrates the sequence of events and communication in the second scenario. Most of the communication sequences in the second scenario are similar to the first scenario. However, additional communication between the agent and the hardware probing process on the adjacent nodes is required. The additional communication sequences enable the carrier agent to select one target node on which a new process can be spawned from the eight adjacent nodes. In summary, a classic version and a failure handling parallel summation algorithm were implemented. The failure handling algorithm incorporates concepts of the agent based failure handling considered in section 2. The algorithms capability of handling single node failure scenarios is demonstrated. The quantitative results obtained from experiments based on the single node failure scenarios of the agent based parallel summation algorithm is reported in the next section. Fig. 5. T sn1 plotted for third level nodes N 13 and N 14 IV. RESULTS The quantitative results obtained from the experiments performed are based on the single node failure scenarios considered in the above section. This section presents the results obtained from both the scenarios. Nodes N 9 N 15 as shown in figure 1 are the computational nodes of the parallel summation algorithm. In the experimental results reported in this section the third level nodes, nodes N 13 and N 14 were only considered. A. Single Node Failure Scenario 1 The time taken by an agent to transfer itself in single node failure scenario 1 from a node predicted to fail onto another adjacent node in the abstracted layer and re-establish all process dependencies for seamless execution, otherwise referred to as T sn1 shown in figure 3 was noted. Thirty different trial runs were performed to gather the statistic. Figure 5 shows the graph that plots T sn1 for nodes N 13 and N 14. The mean of T sn1 is calculated as seconds and is shown in figure 5 as a red axis line. B. Single Node Failure Scenario 2 The time taken by an agent to transfer itself in single node failure scenario 2 from a node predicted to fail onto another adjacent node in the abstracted layer and re-establish all process dependencies for seamless execution, otherwise referred to as T sn2 shown in figure 4 was noted. Thirty different trial runs were performed to gather the statistic. Figure 5 shows the graph that plots T sn2 for nodes N 13 and N 14. The mean of T sn2 is calculated as seconds and is shown in figure 6 as a red axis line. It is noted that the mean time for the single node failure scenario 2 requires greater time. This is due to additional communication sequences to gather sensory information that aids decision making concerning which target node can be Fig. 6. T sn2 plotted for third level nodes N 13 and N 14 used to spawn a new process. The additional time is calculated as T x = T sn2 T sn1 and is obtained as seconds. The mean time taken by an agent to transfer itself in a realistic scenario (single node failure scenario 2) from a node predicted to fail onto another adjacent node in the abstracted layer and re-establish all process dependencies for seamless execution or in other words the mean time taken for reinstating execution after a predicted node failure is noted as seconds. If traditional checkpointing (checkpointing only when the failure is predicted) with human adminstration was employed or incremental checkpointing (periodic checkpointing so that a process does not need to restart from the beginning), reinstating execution would atleast be in the order of minutes. This brief comparison reveals that the agent based approach is effective than traditional failure handling methods. In short, though preliminary results obtained through simple experiments are presented, the agent based approach proposed in this paper is promising and paves a path towards being able to handle faults more efficiently than traditional fault handling

6 approaches in distributed parallel computing systems. V. CONCLUSION In this paper, handling single node failures for parallel summation algorithms in computer clusters has been considered. An agent based approach has been proposed in which a task to be executed is decomposed to sub-tasks and mapped onto agents that traverse computing nodes. The agents intercommunicate across computing nodes to share information during the event of a predicted node failure. Two single node failure scenarios are considered. It is also observed that implementing a realistic scenario for single node failures require additional communication sequences between the agent and hardware probing processes at the expense of time. The Message Passing Interface has been employed for implementing the proposed approach. Quantitative results obtained from experiments reveal that the agent based approach can handle failures more efficiently than traditional failure handling approaches. Future work will aim to extend the agent based approach for multiple node failures. More statistical results will be gathered to compare the efficiency of the proposed approach with existing and traditional failure handling approaches. Efforts will also me made to implement the approach on other largescale parallel computing systems. [11] S.Summiya, K. Ijaz, U. Manzoor and A. A. Shahid, A Fault Tolerant Infrastructure for Mobile Agents, Proceedings of the International Conference on Computational Intelligence for Modelling Control and Automation, [12] D. E. Comer, Internetworking with TCP/IP, Volume 1: Principles, Protocols, and Architecture, Prentice Hall, [13] M. Wooldridge, An Introduction to Multi-Agent Systems, Second Edition, John Wiley & Sons, [14] D. Weyns, H. Van Dyke Parunak and F. Michel, Environments for Multi- Agent Systems, Lecture Notes in Artifcial Intelligence 3374, Springer, [15] M. J. Quinn, Parallel Computing Theory and Practice, McGraw-Hill Inc., [16] J. D. Sloan, High Performance Linux Cluster with OSCAR, Rocks, openmosix & MPI, O Reilly, [17] W. Gropp, E. Lusk and A. Skjullum, Using MPI-2: Advanced Features of the Message Passing Interface, MIT Press, [18] Center for Advanced Computing and Emerging Technologies (ACET) website: [19] High Performance Computing at ACET website: [20] OpenMPI website: [21] E. Gabriel, G. E. Fagg, G. Bosilca, T. Angskun, J. Dongarra, J. M. Squyres, V. Sahay, P. Kambadur, B. Barrett, A. Lumsdaine, R. H. Castain, D. J. Daniel, R. L. Graham, T. S. Woodall, Open MPI: Goals, Concept, and Design of a Next Generation MPI Implementation, Proceedings of the 11th European PVM/MPI Users Group Meeting, Budapest, Hungary, 2004, pp [22] MPI Tutorial: report.html REFERENCES [1] K. A. Hummel and G. Jelleschitz, A Robust Decentralized Job Scheduling Approach for Mobile Peers in Ad-hoc Grids, Proceedings of the 7th IEEE International Symposium on Cluster Computing and Grid, 2007, pp [2] C. Engelmann, G. R. Vallee, T. Naughton and S. L. Scott, Proactive Fault Tolerance using Preemptive Migration, Proceedings of the 17th Euromicro International Conference on Parallel, Distributed and Networkbased Processing, 2009, pp [3] B. Eckart, X. Chen, X. He and S. L. Scott, Failure Prediction Models for Proactive Fault Tolerance within Storage Systems, Proceedings of the IEEE International Symposium on Modelling, Analysis and Simulation of Computers and Telecommunication Systems, 2008, pp [4] A. F. Iskander and A. A. Younis, A Proactive Fault tolerance Management Algorithm for Mobile Ad Hoc Networks, Proceedings of the 4th IEEE Consumer Communications and Networking Conference, 2007, pp [5] J. P. Walters and V. Chaudhary, Replication-Based Fault Tolerance for MPI Applications, IEEE Transactions on Parallel and Distributed Systems, Vol. 20, No. 7, July 2009, pp [6] X. Yang, Y. Du, P. Wang, H. Fu and J. Jia, FTPA: Supporting Fault- Tolerant Parallel Computing through Parallel Recomputing, IEEE Transactions on Parallel and Distributed Systems, Vol. 20, Issue 10, October 2009, pp [7] P. Tichy, P. Slechta, R. J. Staron, F. P. Maturana and K. H. Hall, Multi-agent Technology or Fault Tolerance and Flexible Control, IEEE Transactions on Systems, Man and Cybernetics, Part C: Application and Reviews, 2006, pp [8] M. J. G. C. Mendes, B. M. S. Santos and J. Sa da Costa, Multi-agent Platform for Fault Tolerant Control Systems, Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, 2007, pp [9] Ad. L. Almeida, S. Aknine, J. -P. Briot and J. Malenfant, Plan-Based Replication for Fault-Tolerant Multi-Agent Systems, Proceedings of the 20th IEEE International Parallel and Distributed Processing Symposium, [10] Z. A. Khan, S. Shahid, H. F. Ahmad, A. Ali and H. Suguri, Decentralized Architecture for Fault Tolerant Multi Agent System, Proceedings of the 7th IEEE International Symposium on Autonomous Decentralized Systems, 2005, pp

Improving the Dynamic Creation of Processes in MPI-2

Improving the Dynamic Creation of Processes in MPI-2 Improving the Dynamic Creation of Processes in MPI-2 Márcia C. Cera, Guilherme P. Pezzi, Elton N. Mathias, Nicolas Maillard, and Philippe O. A. Navaux Universidade Federal do Rio Grande do Sul, Instituto

More information

SELF-HEALING NETWORK FOR SCALABLE FAULT TOLERANT RUNTIME ENVIRONMENTS

SELF-HEALING NETWORK FOR SCALABLE FAULT TOLERANT RUNTIME ENVIRONMENTS SELF-HEALING NETWORK FOR SCALABLE FAULT TOLERANT RUNTIME ENVIRONMENTS Thara Angskun, Graham Fagg, George Bosilca, Jelena Pješivac Grbović, and Jack Dongarra,2,3 University of Tennessee, 2 Oak Ridge National

More information

Implementing a Hardware-Based Barrier in Open MPI

Implementing a Hardware-Based Barrier in Open MPI Implementing a Hardware-Based Barrier in Open MPI - A Case Study - Torsten Hoefler 1, Jeffrey M. Squyres 2, Torsten Mehlan 1 Frank Mietke 1 and Wolfgang Rehm 1 1 Technical University of Chemnitz 2 Open

More information

A Survey on Resource Allocation policies in Mobile ad-hoc Computational Network

A Survey on Resource Allocation policies in Mobile ad-hoc Computational Network A Survey on policies in Mobile ad-hoc Computational S. Kamble 1, A. Savyanavar 2 1PG Scholar, Department of Computer Engineering, MIT College of Engineering, Pune, Maharashtra, India 2Associate Professor,

More information

Scalable Fault Tolerant Protocol for Parallel Runtime Environments

Scalable Fault Tolerant Protocol for Parallel Runtime Environments Scalable Fault Tolerant Protocol for Parallel Runtime Environments Thara Angskun, Graham E. Fagg, George Bosilca, Jelena Pješivac Grbović, and Jack J. Dongarra Dept. of Computer Science, 1122 Volunteer

More information

ANALYZING THE EFFICIENCY OF PROGRAM THROUGH VARIOUS OOAD METRICS

ANALYZING THE EFFICIENCY OF PROGRAM THROUGH VARIOUS OOAD METRICS ANALYZING THE EFFICIENCY OF PROGRAM THROUGH VARIOUS OOAD METRICS MR. S. PASUPATHY 1 AND DR. R. BHAVANI 2 1 Associate Professor, Dept. of CSE, FEAT, Annamalai University, Tamil Nadu, India. 2 Professor,

More information

Scalable Middleware Environment for Agent-Based Internet Applications]

Scalable Middleware Environment for Agent-Based Internet Applications] Scalable Middleware Environment for Agent-Based Internet Applications] Benno J. Overeinder and Frances M.T. Brazier Department of Computer Science, Vrije Universiteit Amsterdam De Boelelaan 1081a, 1081

More information

Scalable Fault Tolerant Protocol for Parallel Runtime Environments

Scalable Fault Tolerant Protocol for Parallel Runtime Environments Scalable Fault Tolerant Protocol for Parallel Runtime Environments Thara Angskun 1, Graham Fagg 1, George Bosilca 1, Jelena Pješivac Grbović 1, and Jack Dongarra 2 1 Department of Computer Science, The

More information

Evaluating Algorithms for Shared File Pointer Operations in MPI I/O

Evaluating Algorithms for Shared File Pointer Operations in MPI I/O Evaluating Algorithms for Shared File Pointer Operations in MPI I/O Ketan Kulkarni and Edgar Gabriel Parallel Software Technologies Laboratory, Department of Computer Science, University of Houston {knkulkarni,gabriel}@cs.uh.edu

More information

TEG: A High-Performance, Scalable, Multi-Network Point-to-Point Communications Methodology

TEG: A High-Performance, Scalable, Multi-Network Point-to-Point Communications Methodology TEG: A High-Performance, Scalable, Multi-Network Point-to-Point Communications Methodology T.S. Woodall 1, R.L. Graham 1, R.H. Castain 1, D.J. Daniel 1, M.W. Sukalski 2, G.E. Fagg 3, E. Gabriel 3, G. Bosilca

More information

Technical Comparison between several representative checkpoint/rollback solutions for MPI programs

Technical Comparison between several representative checkpoint/rollback solutions for MPI programs Technical Comparison between several representative checkpoint/rollback solutions for MPI programs Yuan Tang Innovative Computing Laboratory Department of Computer Science University of Tennessee Knoxville,

More information

Fault Tolerant Domain Decomposition for Parabolic Problems

Fault Tolerant Domain Decomposition for Parabolic Problems Fault Tolerant Domain Decomposition for Parabolic Problems Marc Garbey and Hatem Ltaief Department of Computer Science, University of Houston, Houston, TX 77204 USA garbey@cs.uh.edu, ltaief@cs.uh.edu 1

More information

OVERHEADS ENHANCEMENT IN MUTIPLE PROCESSING SYSTEMS BY ANURAG REDDY GANKAT KARTHIK REDDY AKKATI

OVERHEADS ENHANCEMENT IN MUTIPLE PROCESSING SYSTEMS BY ANURAG REDDY GANKAT KARTHIK REDDY AKKATI CMPE 655- MULTIPLE PROCESSOR SYSTEMS OVERHEADS ENHANCEMENT IN MUTIPLE PROCESSING SYSTEMS BY ANURAG REDDY GANKAT KARTHIK REDDY AKKATI What is MULTI PROCESSING?? Multiprocessing is the coordinated processing

More information

ADAPTIVE AND DYNAMIC LOAD BALANCING METHODOLOGIES FOR DISTRIBUTED ENVIRONMENT

ADAPTIVE AND DYNAMIC LOAD BALANCING METHODOLOGIES FOR DISTRIBUTED ENVIRONMENT ADAPTIVE AND DYNAMIC LOAD BALANCING METHODOLOGIES FOR DISTRIBUTED ENVIRONMENT PhD Summary DOCTORATE OF PHILOSOPHY IN COMPUTER SCIENCE & ENGINEERING By Sandip Kumar Goyal (09-PhD-052) Under the Supervision

More information

MPI Collective Algorithm Selection and Quadtree Encoding

MPI Collective Algorithm Selection and Quadtree Encoding MPI Collective Algorithm Selection and Quadtree Encoding Jelena Pješivac Grbović, Graham E. Fagg, Thara Angskun, George Bosilca, and Jack J. Dongarra Innovative Computing Laboratory, University of Tennessee

More information

Evaluating Sparse Data Storage Techniques for MPI Groups and Communicators

Evaluating Sparse Data Storage Techniques for MPI Groups and Communicators Evaluating Sparse Data Storage Techniques for MPI Groups and Communicators Mohamad Chaarawi and Edgar Gabriel Parallel Software Technologies Laboratory, Department of Computer Science, University of Houston,

More information

Similarities and Differences Between Parallel Systems and Distributed Systems

Similarities and Differences Between Parallel Systems and Distributed Systems Similarities and Differences Between Parallel Systems and Distributed Systems Pulasthi Wickramasinghe, Geoffrey Fox School of Informatics and Computing,Indiana University, Bloomington, IN 47408, USA In

More information

Multicast can be implemented here

Multicast can be implemented here MPI Collective Operations over IP Multicast? Hsiang Ann Chen, Yvette O. Carrasco, and Amy W. Apon Computer Science and Computer Engineering University of Arkansas Fayetteville, Arkansas, U.S.A fhachen,yochoa,aapong@comp.uark.edu

More information

PARALLEL PROGRAM EXECUTION SUPPORT IN THE JGRID SYSTEM

PARALLEL PROGRAM EXECUTION SUPPORT IN THE JGRID SYSTEM PARALLEL PROGRAM EXECUTION SUPPORT IN THE JGRID SYSTEM Szabolcs Pota 1, Gergely Sipos 2, Zoltan Juhasz 1,3 and Peter Kacsuk 2 1 Department of Information Systems, University of Veszprem, Hungary 2 Laboratory

More information

GEMOM Genetic Message Oriented Secure Middleware Significant and Measureable Progress beyond the State of the Art

GEMOM Genetic Message Oriented Secure Middleware Significant and Measureable Progress beyond the State of the Art GEMOM Genetic Message Oriented Secure Middleware Significant and Measureable Progress beyond the State of the Art Habtamu Abie, Ilesh Dattani,, Milan Novkovic,, John Bigham,, Shaun Topham,, and Reijo Savola

More information

Group Management Schemes for Implementing MPI Collective Communication over IP Multicast

Group Management Schemes for Implementing MPI Collective Communication over IP Multicast Group Management Schemes for Implementing MPI Collective Communication over IP Multicast Xin Yuan Scott Daniels Ahmad Faraj Amit Karwande Department of Computer Science, Florida State University, Tallahassee,

More information

Runtime Optimization of Application Level Communication Patterns

Runtime Optimization of Application Level Communication Patterns Runtime Optimization of Application Level Communication Patterns Edgar Gabriel and Shuo Huang Department of Computer Science, University of Houston, Houston, TX, USA {gabriel, shhuang}@cs.uh.edu Abstract

More information

Coupling DDT and Marmot for Debugging of MPI Applications

Coupling DDT and Marmot for Debugging of MPI Applications Coupling DDT and Marmot for Debugging of MPI Applications Bettina Krammer 1, Valentin Himmler 1, and David Lecomber 2 1 HLRS - High Performance Computing Center Stuttgart, Nobelstrasse 19, 70569 Stuttgart,

More information

A Scalable Parallel Genetic Algorithm for X-ray Spectroscopic Analysis

A Scalable Parallel Genetic Algorithm for X-ray Spectroscopic Analysis A Scalable Parallel Genetic Algorithm for X-ray Spectroscopic Analysis Kai Xu Dept. of Computer Science and Engineering University of Nevada, Reno Reno, NV, 89557 xukai@cs.unr.edu Sushil J. Louis Dept.

More information

Navigation of Multiple Mobile Robots Using Swarm Intelligence

Navigation of Multiple Mobile Robots Using Swarm Intelligence Navigation of Multiple Mobile Robots Using Swarm Intelligence Dayal R. Parhi National Institute of Technology, Rourkela, India E-mail: dayalparhi@yahoo.com Jayanta Kumar Pothal National Institute of Technology,

More information

Integrating Heterogeneous Computing Techniques into the Daliuge Execution Framework

Integrating Heterogeneous Computing Techniques into the Daliuge Execution Framework Integrating Heterogeneous Computing Techniques into the Daliuge Execution Framework Feng Wang, Toby Potter, Xavier Simmons, Shifan Zuo, Jiangying Gan May 15, 2017 Abstract The ability to run MPI-based

More information

Synchronization and Dissemination in Self-Organizing Communication Networks

Synchronization and Dissemination in Self-Organizing Communication Networks Synchronization and Dissemination in Self-Organizing Communication Networks Christian Bettstetter Networked and Embedded Systems University of Klagenfurt Lakeside Labs GmbH Talk at the Vrije Universiteit

More information

Parallel Sorting with Minimal Data

Parallel Sorting with Minimal Data Parallel Sorting with Minimal Data Christian Siebert 1,2 and Felix Wolf 1,2,3 1 German Research School for Simulation Sciences, 52062 Aachen, Germany 2 RWTH Aachen University, Computer Science Department,

More information

Performance Analysis and Optimal Utilization of Inter-Process Communications on Commodity Clusters

Performance Analysis and Optimal Utilization of Inter-Process Communications on Commodity Clusters Performance Analysis and Optimal Utilization of Inter-Process Communications on Commodity Yili TSENG Department of Computer Systems Technology North Carolina A & T State University Greensboro, NC 27411,

More information

A Capabilities Based Communication Model for High-Performance Distributed Applications: The Open HPC++ Approach

A Capabilities Based Communication Model for High-Performance Distributed Applications: The Open HPC++ Approach A Capabilities Based Communication Model for High-Performance Distributed Applications: The Open HPC++ Approach Shridhar Diwan, Dennis Gannon Department of Computer Science Indiana University Bloomington,

More information

MPI History. MPI versions MPI-2 MPICH2

MPI History. MPI versions MPI-2 MPICH2 MPI versions MPI History Standardization started (1992) MPI-1 completed (1.0) (May 1994) Clarifications (1.1) (June 1995) MPI-2 (started: 1995, finished: 1997) MPI-2 book 1999 MPICH 1.2.4 partial implemention

More information

MPI versions. MPI History

MPI versions. MPI History MPI versions MPI History Standardization started (1992) MPI-1 completed (1.0) (May 1994) Clarifications (1.1) (June 1995) MPI-2 (started: 1995, finished: 1997) MPI-2 book 1999 MPICH 1.2.4 partial implemention

More information

An agent-based peer-to-peer grid computing architecture

An agent-based peer-to-peer grid computing architecture University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2005 An agent-based peer-to-peer grid computing architecture J. Tang University

More information

Co-operative Scheduled Energy Aware Load-Balancing technique for an Efficient Computational Cloud

Co-operative Scheduled Energy Aware Load-Balancing technique for an Efficient Computational Cloud 571 Co-operative Scheduled Energy Aware Load-Balancing technique for an Efficient Computational Cloud T.R.V. Anandharajan 1, Dr. M.A. Bhagyaveni 2 1 Research Scholar, Department of Electronics and Communication,

More information

6. NEURAL NETWORK BASED PATH PLANNING ALGORITHM 6.1 INTRODUCTION

6. NEURAL NETWORK BASED PATH PLANNING ALGORITHM 6.1 INTRODUCTION 6 NEURAL NETWORK BASED PATH PLANNING ALGORITHM 61 INTRODUCTION In previous chapters path planning algorithms such as trigonometry based path planning algorithm and direction based path planning algorithm

More information

Parallel Evaluation of Hopfield Neural Networks

Parallel Evaluation of Hopfield Neural Networks Parallel Evaluation of Hopfield Neural Networks Antoine Eiche, Daniel Chillet, Sebastien Pillement and Olivier Sentieys University of Rennes I / IRISA / INRIA 6 rue de Kerampont, BP 818 2232 LANNION,FRANCE

More information

Hierarchical Addressing and Routing Mechanisms for Distributed Applications over Heterogeneous Networks

Hierarchical Addressing and Routing Mechanisms for Distributed Applications over Heterogeneous Networks Hierarchical Addressing and Routing Mechanisms for Distributed Applications over Heterogeneous Networks Damien Magoni Université Louis Pasteur LSIIT magoni@dpt-info.u-strasbg.fr Abstract. Although distributed

More information

A Parallel Evolutionary Algorithm for Discovery of Decision Rules

A Parallel Evolutionary Algorithm for Discovery of Decision Rules A Parallel Evolutionary Algorithm for Discovery of Decision Rules Wojciech Kwedlo Faculty of Computer Science Technical University of Bia lystok Wiejska 45a, 15-351 Bia lystok, Poland wkwedlo@ii.pb.bialystok.pl

More information

Job Re-Packing for Enhancing the Performance of Gang Scheduling

Job Re-Packing for Enhancing the Performance of Gang Scheduling Job Re-Packing for Enhancing the Performance of Gang Scheduling B. B. Zhou 1, R. P. Brent 2, C. W. Johnson 3, and D. Walsh 3 1 Computer Sciences Laboratory, Australian National University, Canberra, ACT

More information

I/O in the Gardens Non-Dedicated Cluster Computing Environment

I/O in the Gardens Non-Dedicated Cluster Computing Environment I/O in the Gardens Non-Dedicated Cluster Computing Environment Paul Roe and Siu Yuen Chan School of Computing Science Queensland University of Technology Australia fp.roe, s.chang@qut.edu.au Abstract Gardens

More information

Chapter 3. Design of Grid Scheduler. 3.1 Introduction

Chapter 3. Design of Grid Scheduler. 3.1 Introduction Chapter 3 Design of Grid Scheduler The scheduler component of the grid is responsible to prepare the job ques for grid resources. The research in design of grid schedulers has given various topologies

More information

Expressing Fault Tolerant Algorithms with MPI-2. William D. Gropp Ewing Lusk

Expressing Fault Tolerant Algorithms with MPI-2. William D. Gropp Ewing Lusk Expressing Fault Tolerant Algorithms with MPI-2 William D. Gropp Ewing Lusk www.mcs.anl.gov/~gropp Overview Myths about MPI and Fault Tolerance Error handling and reporting Goal of Fault Tolerance Run

More information

A Reliable Broadcast System

A Reliable Broadcast System A Reliable Broadcast System Yuchen Dai, Xiayi Huang, Diansan Zhou Department of Computer Sciences and Engineering Santa Clara University December 10 2013 Table of Contents 2 Introduction......3 2.1 Objective...3

More information

Influence of the Progress Engine on the Performance of Asynchronous Communication Libraries

Influence of the Progress Engine on the Performance of Asynchronous Communication Libraries Influence of the Progress Engine on the Performance of Asynchronous Communication Libraries Edgar Gabriel Department of Computer Science University of Houston Houston, TX, 77204, USA http://www.cs.uh.edu

More information

BUILDING A HIGHLY SCALABLE MPI RUNTIME LIBRARY ON GRID USING HIERARCHICAL VIRTUAL CLUSTER APPROACH

BUILDING A HIGHLY SCALABLE MPI RUNTIME LIBRARY ON GRID USING HIERARCHICAL VIRTUAL CLUSTER APPROACH BUILDING A HIGHLY SCALABLE MPI RUNTIME LIBRARY ON GRID USING HIERARCHICAL VIRTUAL CLUSTER APPROACH Theewara Vorakosit and Putchong Uthayopas High Performance Computing and Networking Center Faculty of

More information

A Component Architecture for LAM/MPI

A Component Architecture for LAM/MPI A Component Architecture for LAM/MPI Jeffrey M. Squyres and Andrew Lumsdaine Open Systems Lab, Indiana University Abstract. To better manage the ever increasing complexity of

More information

Study of Load Balancing Schemes over a Video on Demand System

Study of Load Balancing Schemes over a Video on Demand System Study of Load Balancing Schemes over a Video on Demand System Priyank Singhal Ashish Chhabria Nupur Bansal Nataasha Raul Research Scholar, Computer Department Abstract: Load balancing algorithms on Video

More information

Scalable, Fault-Tolerant Membership for MPI Tasks on HPC Systems

Scalable, Fault-Tolerant Membership for MPI Tasks on HPC Systems fastos.org/molar Scalable, Fault-Tolerant Membership for MPI Tasks on HPC Systems Jyothish Varma 1, Chao Wang 1, Frank Mueller 1, Christian Engelmann, Stephen L. Scott 1 North Carolina State University,

More information

Virtual Machine (VM) Earlier Failure Prediction Algorithm

Virtual Machine (VM) Earlier Failure Prediction Algorithm Virtual Machine (VM) Earlier Failure Prediction Algorithm Shaima a Ghazi Research Scholar, Department of Computer Science, Jain University, #1/1-1, Atria Towers, Palace Road, Bangalore, Karnataka, India.

More information

Mobile Element Scheduling for Efficient Data Collection in Wireless Sensor Networks: A Survey

Mobile Element Scheduling for Efficient Data Collection in Wireless Sensor Networks: A Survey Journal of Computer Science 7 (1): 114-119, 2011 ISSN 1549-3636 2011 Science Publications Mobile Element Scheduling for Efficient Data Collection in Wireless Sensor Networks: A Survey K. Indra Gandhi and

More information

Chapter 3: AIS Enhancements Through Information Technology and Networks

Chapter 3: AIS Enhancements Through Information Technology and Networks Accounting Information Systems: Essential Concepts and Applications Fourth Edition by Wilkinson, Cerullo, Raval, and Wong-On-Wing Chapter 3: AIS Enhancements Through Information Technology and Networks

More information

CHAPTER 7 CONCLUSION AND FUTURE WORK

CHAPTER 7 CONCLUSION AND FUTURE WORK CHAPTER 7 CONCLUSION AND FUTURE WORK 7.1 Conclusion Data pre-processing is very important in data mining process. Certain data cleaning techniques usually are not applicable to all kinds of data. Deduplication

More information

Network Survivability Simulation of a Commercially Deployed Dynamic Routing System Protocol

Network Survivability Simulation of a Commercially Deployed Dynamic Routing System Protocol Network Survivability Simulation of a Commercially Deployed Dynamic Routing System Protocol Abdur Chowdhury 1,, Ophir Frieder 1, Paul Luse, Peng-Jun Wan 1 {abdur, wan, ophir}@cs.iit.edu, pluse@iitri.org

More information

Adding Low-Cost Hardware Barrier Support to Small Commodity Clusters

Adding Low-Cost Hardware Barrier Support to Small Commodity Clusters Adding Low-Cost Hardware Barrier Support to Small Commodity Clusters Torsten Hoefler, Torsten Mehlan, Frank Mietke and Wolfgang Rehm {htor,tome,mief,rehm}@cs.tu-chemnitz.de 17th January 2006 Abstract The

More information

Reducing the Number of Test Cases for Performance Evaluation of Components

Reducing the Number of Test Cases for Performance Evaluation of Components Reducing the Number of Test Cases for Performance Evaluation of Components João W. Cangussu Kendra Cooper Eric Wong Department of Computer Science University of Texas at Dallas Richardson-TX 75-, USA cangussu,kcooper,ewong

More information

Finding Optimal Tour Length of Mobile Agent in Wireless Sensor Network

Finding Optimal Tour Length of Mobile Agent in Wireless Sensor Network Conference on Advances in Communication and Control Systems 2013 (CAC2S 2013) Finding Optimal Tour Length of Mobile Agent in Wireless Sensor Network Anil Kumar Mahto anil.fiem16@gmail.com Ajay Prasad Department

More information

Energy efficient mapping of virtual machines

Energy efficient mapping of virtual machines GreenDays@Lille Energy efficient mapping of virtual machines Violaine Villebonnet Thursday 28th November 2013 Supervisor : Georges DA COSTA 2 Current approaches for energy savings in cloud Several actions

More information

Scalable Hybrid Search on Distributed Databases

Scalable Hybrid Search on Distributed Databases Scalable Hybrid Search on Distributed Databases Jungkee Kim 1,2 and Geoffrey Fox 2 1 Department of Computer Science, Florida State University, Tallahassee FL 32306, U.S.A., jungkkim@cs.fsu.edu, 2 Community

More information

Analysis of Cluster-Based Energy-Dynamic Routing Protocols in WSN

Analysis of Cluster-Based Energy-Dynamic Routing Protocols in WSN Analysis of Cluster-Based Energy-Dynamic Routing Protocols in WSN Mr. V. Narsing Rao 1, Dr.K.Bhargavi 2 1,2 Asst. Professor in CSE Dept., Sphoorthy Engineering College, Hyderabad Abstract- Wireless Sensor

More information

Application of Autonomous Mobile Agents to Provide Security in Wireless Networks

Application of Autonomous Mobile Agents to Provide Security in Wireless Networks Application of Autonomous Mobile Agents to Provide Security in Wireless Networks ¹Odhiambo MO, ²Aderemi Lawal Department of Electrical and Mining Engineering, University of South Africa, Pretoria, South

More information

Introducing Robotics Vision System to a Manufacturing Robotics Course

Introducing Robotics Vision System to a Manufacturing Robotics Course Paper ID #16241 Introducing Robotics Vision System to a Manufacturing Robotics Course Dr. Yuqiu You, Ohio University c American Society for Engineering Education, 2016 Introducing Robotics Vision System

More information

A COMPARISON OF REACTIVE ROUTING PROTOCOLS DSR, AODV AND TORA IN MANET

A COMPARISON OF REACTIVE ROUTING PROTOCOLS DSR, AODV AND TORA IN MANET ISSN: 2278 1323 All Rights Reserved 2016 IJARCET 296 A COMPARISON OF REACTIVE ROUTING PROTOCOLS DSR, AODV AND TORA IN MANET Dr. R. Shanmugavadivu 1, B. Chitra 2 1 Assistant Professor, Department of Computer

More information

Factors Affecting the Performance of Ad Hoc Networks

Factors Affecting the Performance of Ad Hoc Networks Factors Affecting the Performance of Ad Hoc Networks Dmitri D. Perkins, Herman D. Hughes, and Charles B. Owen Department of Computer Science and Engineering Michigan State University East Lansing, MI 88-6

More information

Flexible collective communication tuning architecture applied to Open MPI

Flexible collective communication tuning architecture applied to Open MPI Flexible collective communication tuning architecture applied to Open MPI Graham E. Fagg 1, Jelena Pjesivac-Grbovic 1, George Bosilca 1, Thara Angskun 1, Jack J. Dongarra 1, and Emmanuel Jeannot 2 1 Dept.

More information

HETEROGENEOUS COMPUTING

HETEROGENEOUS COMPUTING HETEROGENEOUS COMPUTING Shoukat Ali, Tracy D. Braun, Howard Jay Siegel, and Anthony A. Maciejewski School of Electrical and Computer Engineering, Purdue University Heterogeneous computing is a set of techniques

More information

Visual Debugging of MPI Applications

Visual Debugging of MPI Applications Visual Debugging of MPI Applications Basile Schaeli 1, Ali Al-Shabibi 1 and Roger D. Hersch 1 1 Ecole Polytechnique Fédérale de Lausanne (EPFL) School of Computer and Communication Sciences CH-1015 Lausanne,

More information

A Study on the Testing of VLSI Systems Using Reduced Power Consumption Methods

A Study on the Testing of VLSI Systems Using Reduced Power Consumption Methods International Journal of Scientific & Engineering Research, Volume 4, Issue 4, April-2013 664 A Study on the Testing of VLSI Systems Using Reduced Power Consumption Methods Debasmita Hazra Abstract- This

More information

Real-time grid computing for financial applications

Real-time grid computing for financial applications CNR-INFM Democritos and EGRID project E-mail: cozzini@democritos.it Riccardo di Meo, Ezio Corso EGRID project ICTP E-mail: {dimeo,ecorso}@egrid.it We describe the porting of a test case financial application

More information

Scheduling Dynamically Spawned Processes in MPI-2

Scheduling Dynamically Spawned Processes in MPI-2 Scheduling Dynamically Spawned Processes in MPI-2 Márcia C. Cera 1, Guilherme P. Pezzi 1, Maurício L. Pilla 2, Nicolas B. Maillard 1, and Philippe O. A. Navaux 1 1 Universidade Federal do Rio Grande do

More information

Wave front Method Based Path Planning Algorithm for Mobile Robots

Wave front Method Based Path Planning Algorithm for Mobile Robots Wave front Method Based Path Planning Algorithm for Mobile Robots Bhavya Ghai 1 and Anupam Shukla 2 ABV- Indian Institute of Information Technology and Management, Gwalior, India 1 bhavyaghai@gmail.com,

More information

A Comparative Study of Load Balancing Algorithms: A Review Paper

A Comparative Study of Load Balancing Algorithms: A Review Paper Available Online at www.ijcsmc.com International Journal of Computer Science and Mobile Computing A Monthly Journal of Computer Science and Information Technology ISSN 2320 088X IMPACT FACTOR: 5.258 IJCSMC,

More information

Image-Space-Parallel Direct Volume Rendering on a Cluster of PCs

Image-Space-Parallel Direct Volume Rendering on a Cluster of PCs Image-Space-Parallel Direct Volume Rendering on a Cluster of PCs B. Barla Cambazoglu and Cevdet Aykanat Bilkent University, Department of Computer Engineering, 06800, Ankara, Turkey {berkant,aykanat}@cs.bilkent.edu.tr

More information

An Architecture For Computational Grids Based On Proxy Servers

An Architecture For Computational Grids Based On Proxy Servers An Architecture For Computational Grids Based On Proxy Servers P. V. C. Costa, S. D. Zorzo, H. C. Guardia {paulocosta,zorzo,helio}@dc.ufscar.br UFSCar Federal University of São Carlos, Brazil Abstract

More information

Mobile Robot Path Planning in Static Environments using Particle Swarm Optimization

Mobile Robot Path Planning in Static Environments using Particle Swarm Optimization Mobile Robot Path Planning in Static Environments using Particle Swarm Optimization M. Shahab Alam, M. Usman Rafique, and M. Umer Khan Abstract Motion planning is a key element of robotics since it empowers

More information

Mobile Wireless Sensor Network enables convergence of ubiquitous sensor services

Mobile Wireless Sensor Network enables convergence of ubiquitous sensor services 1 2005 Nokia V1-Filename.ppt / yyyy-mm-dd / Initials Mobile Wireless Sensor Network enables convergence of ubiquitous sensor services Dr. Jian Ma, Principal Scientist Nokia Research Center, Beijing 2 2005

More information

Early Measurements of a Cluster-based Architecture for P2P Systems

Early Measurements of a Cluster-based Architecture for P2P Systems Early Measurements of a Cluster-based Architecture for P2P Systems Balachander Krishnamurthy, Jia Wang, Yinglian Xie I. INTRODUCTION Peer-to-peer applications such as Napster [4], Freenet [1], and Gnutella

More information

Navpreet Singh INTRODUCTION TO COMPUTER NETWORKS. Computer Centre Indian Institute of Technology Kanpur Kanpur INDIA

Navpreet Singh INTRODUCTION TO COMPUTER NETWORKS. Computer Centre Indian Institute of Technology Kanpur Kanpur INDIA INTRODUCTION TO COMPUTER NETWORKS Navpreet Singh Computer Centre Indian Institute of Technology Kanpur Kanpur INDIA (Ph : 2597371, Email : navi@iitk.ac.in) About Myself About Myself I am Network Manager

More information

High Performance Multithreaded Model for Stream Cipher

High Performance Multithreaded Model for Stream Cipher 228 IJCSNS International Journal of Computer Science and Network Security, VOL.8 No.3, March 2008 High Performance Multithreaded Model for Stream Cipher Khaled M. Suwais and Azman Bin Samsudin School of

More information

A Novel Approach to Planar Mechanism Synthesis Using HEEDS

A Novel Approach to Planar Mechanism Synthesis Using HEEDS AB-2033 Rev. 04.10 A Novel Approach to Planar Mechanism Synthesis Using HEEDS John Oliva and Erik Goodman Michigan State University Introduction The problem of mechanism synthesis (or design) is deceptively

More information

Performance Evaluation of National Knowledge Network Connectivity

Performance Evaluation of National Knowledge Network Connectivity International Journal of Computer Applications (975 888) Performance Evaluation of National Knowledge Network Connectivity Vipin Saxena, PhD. Department of Computer Science B.B. Ambedkar University (A

More information

Analysis of the Component Architecture Overhead in Open MPI

Analysis of the Component Architecture Overhead in Open MPI Analysis of the Component Architecture Overhead in Open MPI B. Barrett 1, J.M. Squyres 1, A. Lumsdaine 1, R.L. Graham 2, G. Bosilca 3 Open Systems Laboratory, Indiana University {brbarret, jsquyres, lums}@osl.iu.edu

More information

Lecture 7: February 10

Lecture 7: February 10 CMPSCI 677 Operating Systems Spring 2016 Lecture 7: February 10 Lecturer: Prashant Shenoy Scribe: Tao Sun 7.1 Server Design Issues 7.1.1 Server Design There are two types of server design choices: Iterative

More information

Network Solution for Achieving Large-Scale, High-Availability VoIP Services

Network Solution for Achieving Large-Scale, High-Availability VoIP Services Network Solution for Achieving Large-Scale, High-Availability VoIP Services V Kiyofumi Shigeno V Hisayuki Sekine V Eiichiro Takahashi (Manuscript received March 17, 2006) VoIP technologies that enable

More information

Networked control methods robust to jitter and their evaluation by inverted pendulum

Networked control methods robust to jitter and their evaluation by inverted pendulum Networked control methods robust to jitter and their evaluation by inverted pendulum Naotoshi Adachi, Ken Aida, Masuhiro Nitta, and Kenji Sugimoto {naotos-a, ken-a, masuhi-n, kenji} @is.aist-nara.ac.jp

More information

Adaptive Power Blurring Techniques to Calculate IC Temperature Profile under Large Temperature Variations

Adaptive Power Blurring Techniques to Calculate IC Temperature Profile under Large Temperature Variations Adaptive Techniques to Calculate IC Temperature Profile under Large Temperature Variations Amirkoushyar Ziabari, Zhixi Bian, Ali Shakouri Baskin School of Engineering, University of California Santa Cruz

More information

Lecture 6: Overlay Networks. CS 598: Advanced Internetworking Matthew Caesar February 15, 2011

Lecture 6: Overlay Networks. CS 598: Advanced Internetworking Matthew Caesar February 15, 2011 Lecture 6: Overlay Networks CS 598: Advanced Internetworking Matthew Caesar February 15, 2011 1 Overlay networks: Motivations Protocol changes in the network happen very slowly Why? Internet is shared

More information

Performance Evaluation of Various Routing Protocols in MANET

Performance Evaluation of Various Routing Protocols in MANET 208 Performance Evaluation of Various Routing Protocols in MANET Jaya Jacob 1,V.Seethalakshmi 2 1 II MECS,Sri Shakthi Institute of Science and Technology, Coimbatore, India 2 Associate Professor-ECE, Sri

More information

2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media,

2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising

More information

Software Engineering: Integration Requirements

Software Engineering: Integration Requirements Software Engineering: Integration Requirements AYAZ ISAZADEH Department of Computer Science Tabriz University Tabriz, IRAN Abstract: - This paper presents a discussion of software integration requirements,

More information

Estimating Fault-Detection and Fail-Over Times for Nested Real-Time CORBA Applications

Estimating Fault-Detection and Fail-Over Times for Nested Real-Time CORBA Applications Estimating Fault-Detection and Fail-Over Times for Nested Real-Time CORBA Applications Sukanya Ratanotayanon (speaker) School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 389 Tel:

More information

Multisync: A multiagent system for ubiquitous file synchronization

Multisync: A multiagent system for ubiquitous file synchronization Multisync: A multiagent system for ubiquitous file synchronization Muaz Niazi Umar Manzoor Kiran Ijaz Summiya Hina Saleem Computer Science Department, National University of Computer & Emerging Sciences-FAST,

More information

Evaluation of Cartesian-based Routing Metrics for Wireless Sensor Networks

Evaluation of Cartesian-based Routing Metrics for Wireless Sensor Networks Evaluation of Cartesian-based Routing Metrics for Wireless Sensor Networks Ayad Salhieh Department of Electrical and Computer Engineering Wayne State University Detroit, MI 48202 ai4874@wayne.edu Loren

More information

Cross Layering in MANETs Design: The MobileMAN Approach

Cross Layering in MANETs Design: The MobileMAN Approach National Research Council - Pisa - Italy Cross Layering in MANETs Design: The MobileMAN Approach Marco Conti Italian National Research Council (CNR) IIT Institute Pisa, Italy marco.conti@iit.cnr.it Slide

More information

A Virtual Laboratory for Study of Algorithms

A Virtual Laboratory for Study of Algorithms A Virtual Laboratory for Study of Algorithms Thomas E. O'Neil and Scott Kerlin Computer Science Department University of North Dakota Grand Forks, ND 58202-9015 oneil@cs.und.edu Abstract Empirical studies

More information

Efficient Location Services Using Hierarchical Topology of Mobile Ad Hoc Networks

Efficient Location Services Using Hierarchical Topology of Mobile Ad Hoc Networks Journal of Software Engineering and Applications, 2012, 5, 14-20 http://dx.doi.org/10.4236/jsea.2012.51003 Published Online January 2012 (http://www.scirp.org/journal/jsea) Efficient Location Services

More information

Approximating Fault-Tolerant Steiner Subgraphs in Heterogeneous Wireless Networks

Approximating Fault-Tolerant Steiner Subgraphs in Heterogeneous Wireless Networks Approximating Fault-Tolerant Steiner Subgraphs in Heterogeneous Wireless Networks Ambreen Shahnaz and Thomas Erlebach Department of Computer Science University of Leicester University Road, Leicester LE1

More information

Developing a Thin and High Performance Implementation of Message Passing Interface 1

Developing a Thin and High Performance Implementation of Message Passing Interface 1 Developing a Thin and High Performance Implementation of Message Passing Interface 1 Theewara Vorakosit and Putchong Uthayopas Parallel Research Group Computer and Network System Research Laboratory Department

More information

MICROARRAY IMAGE SEGMENTATION USING CLUSTERING METHODS

MICROARRAY IMAGE SEGMENTATION USING CLUSTERING METHODS Mathematical and Computational Applications, Vol. 5, No. 2, pp. 240-247, 200. Association for Scientific Research MICROARRAY IMAGE SEGMENTATION USING CLUSTERING METHODS Volkan Uslan and Đhsan Ömür Bucak

More information

High Performance Computing Prof. Matthew Jacob Department of Computer Science and Automation Indian Institute of Science, Bangalore

High Performance Computing Prof. Matthew Jacob Department of Computer Science and Automation Indian Institute of Science, Bangalore High Performance Computing Prof. Matthew Jacob Department of Computer Science and Automation Indian Institute of Science, Bangalore Module No # 09 Lecture No # 40 This is lecture forty of the course on

More information

EATSAL: An Energy Aware Task Scheduling Algorithm for Hybrid Networks

EATSAL: An Energy Aware Task Scheduling Algorithm for Hybrid Networks EATSAL: An Energy Aware Task Scheduling Algorithm for Hybrid Networks F. Hussain, A. Akram, M. Zafrullah Abstract The widespread popularity of mobile computing devices, such as Laptops, handheld devices

More information