Key Terms - MinMin, MaxMin, Sufferage, Task Scheduling, Standard Deviation, Load Balancing.

Similar documents
Keywords Cloud Computing, Service Level Agreements (SLA), CloudSim, Monitoring & Controlling SLA Agent, JADE

Laboratory Exercise 6

Distributed Packet Processing Architecture with Reconfigurable Hardware Accelerators for 100Gbps Forwarding Performance on Virtualized Edge Router

DAROS: Distributed User-Server Assignment And Replication For Online Social Networking Applications

Refining SIRAP with a Dedicated Resource Ceiling for Self-Blocking

Universität Augsburg. Institut für Informatik. Approximating Optimal Visual Sensor Placement. E. Hörster, R. Lienhart.

Laboratory Exercise 2

MAT 155: Describing, Exploring, and Comparing Data Page 1 of NotesCh2-3.doc

AUTOMATIC TEST CASE GENERATION USING UML MODELS

Lecture 14: Minimum Spanning Tree I

Analyzing Hydra Historical Statistics Part 2

Routing Definition 4.1

A Practical Model for Minimizing Waiting Time in a Transit Network

Laboratory Exercise 6

(12) Patent Application Publication (10) Pub. No.: US 2003/ A1

Laboratory Exercise 6

ISSN: (Online) Volume 3, Issue 4, April 2015 International Journal of Advance Research in Computer Science and Management Studies

Laboratory Exercise 6

An efficient resource allocation algorithm for OFDMA cooperative relay networks with fairness and QoS guaranteed

Semi-Distributed Load Balancing for Massively Parallel Multicomputer Systems

Hassan Ghaziri AUB, OSB Beirut, Lebanon Key words Competitive self-organizing maps, Meta-heuristics, Vehicle routing problem,

Image authentication and tamper detection using fragile watermarking in spatial domain

SLA Adaptation for Service Overlay Networks

UC Berkeley International Conference on GIScience Short Paper Proceedings

A SIMPLE IMPERATIVE LANGUAGE THE STORE FUNCTION NON-TERMINATING COMMANDS

1 The secretary problem

Increasing Throughput and Reducing Delay in Wireless Sensor Networks Using Interference Alignment

See chapter 8 in the textbook. Dr Muhammad Al Salamah, Industrial Engineering, KFUPM

A Load Balancing Model based on Load-aware for Distributed Controllers. Fengjun Shang, Wenjuan Gong

Topics. Lecture 37: Global Optimization. Issues. A Simple Example: Copy Propagation X := 3 B > 0 Y := 0 X := 4 Y := Z + W A := 2 * 3X

/06/$ IEEE 364

A METHOD OF REAL-TIME NURBS INTERPOLATION WITH CONFINED CHORD ERROR FOR CNC SYSTEMS

Performance of a Robust Filter-based Approach for Contour Detection in Wireless Sensor Networks

Contents. shortest paths. Notation. Shortest path problem. Applications. Algorithms and Networks 2010/2011. In the entire course:

The Association of System Performance Professionals

Multi-Target Tracking In Clutter

Laboratory Exercise 2

(12) Patent Application Publication (10) Pub. No.: US 2011/ A1

Parallel Approaches for Intervals Analysis of Variable Statistics in Large and Sparse Linear Equations with RHS Ranges

Static Batch Mode Heuristic Algorithm for Mapping Independent Tasks in Computational Grid

Planning of scooping position and approach path for loading operation by wheel loader

CS 467/567: Divide and Conquer on the PRAM

DWH Performance Tuning For Better Reporting

3D SMAP Algorithm. April 11, 2012

Evolution of Non-Deterministic Incremental Algorithms. Hugues Juille. Volen Center for Complex Systems. Brandeis University. Waltham, MA

Connected Placement of Disaster Shelters in Modern Cities

A Multi-objective Genetic Algorithm for Reliability Optimization Problem

Diverse: Application-Layer Service Differentiation in Peer-to-Peer Communications

Service and Network Management Interworking in Future Wireless Systems

How to Select Measurement Points in Access Point Localization

Compressed Sensing Image Processing Based on Stagewise Orthogonal Matching Pursuit

An Approach to a Test Oracle for XML Query Testing

Audio-Visual Voice Command Recognition in Noisy Conditions

Computer Arithmetic Homework Solutions. 1 An adder for graphics. 2 Partitioned adder. 3 HDL implementation of a partitioned adder

mapping reult. Our experiment have revealed that for many popular tream application, uch a networking and multimedia application, the number of VC nee

Frequency Table Computation on Dataflow Architecture

[N309] Feedforward Active Noise Control Systems with Online Secondary Path Modeling. Muhammad Tahir Akhtar, Masahide Abe, and Masayuki Kawamata

Motion Control (wheeled robots)

Lecture Outline. Global flow analysis. Global Optimization. Global constant propagation. Liveness analysis. Local Optimization. Global Optimization

Stochastic Search and Graph Techniques for MCM Path Planning Christine D. Piatko, Christopher P. Diehl, Paul McNamee, Cheryl Resch and I-Jeng Wang

else end while End References

Dynamically Reconfigurable Neuron Architecture for the Implementation of Self- Organizing Learning Array

Modelling the impact of cyber attacks on the traffic control centre of an urban automobile transport system by means of enhanced cybersecurity

Correlation Models for Shadow Fading Simulation

A Novel Feature Line Segment Approach for Pattern Classification

Architecture and grid application of cluster computing system

Floating Point CORDIC Based Power Operation

Building a Compact On-line MRF Recognizer for Large Character Set using Structured Dictionary Representation and Vector Quantization Technique

Research Article Longest Path Reroute to Optimize the Optical Multicast Routing in Sparse Splitting WDM Networks

A Basic Prototype for Enterprise Level Project Management

CENTER-POINT MODEL OF DEFORMABLE SURFACE

ETSI TS V ( )

A New Approach to Pipeline FFT Processor

Markov Random Fields in Image Segmentation

International Journal of Engineering Research & Technology (IJERT) ISSN: Vol. 2 Issue 5, May

New Structural Decomposition Techniques for Constraint Satisfaction Problems

Advanced Encryption Standard and Modes of Operation

QoS Guided Min-Mean Task Scheduling Algorithm for Scheduling Dr.G.K.Kamalam

Keywords: Defect detection, linear phased array transducer, parameter optimization, phased array ultrasonic B-mode imaging testing.

Cutting Stock by Iterated Matching. Andreas Fritsch, Oliver Vornberger. University of Osnabruck. D Osnabruck.

Distributed Partial Information Management (DPIM) Schemes for Survivable Networks - Part II

Maneuverable Relays to Improve Energy Efficiency in Sensor Networks

Course Project: Adders, Subtractors, and Multipliers a

Spring 2012 EE457 Instructor: Gandhi Puvvada

On combining Learning Vector Quantization and the Bayesian classifiers for natural textured images

Shortest Path Routing in Arbitrary Networks

Chapter 13 Non Sampling Errors

Minimum congestion spanning trees in bipartite and random graphs

A Linear Interpolation-Based Algorithm for Path Planning and Replanning on Girds *

Aalborg Universitet. Published in: Proceedings of the Working Conference on Advanced Visual Interfaces

A CLUSTERING-BASED HYBRID REPLICA CONTROL PROTOCOL FOR HIGH AVAILABILITY IN GRID ENVIRONMENT

Modeling of underwater vehicle s dynamics

Karen L. Collins. Wesleyan University. Middletown, CT and. Mark Hovey MIT. Cambridge, MA Abstract

999 Computer System Network. (12) Patent Application Publication (10) Pub. No.: US 2006/ A1. (19) United States

The Data Locality of Work Stealing

Implementation of a momentum-based distance metric for motion graphs. Student: Alessandro Di Domenico (st.no ), Supervisor: Nicolas Pronost

np vp cost = 0 cost = c np vp cost = c I replacing term cost = c+c n cost = c * Error detection Error correction pron det pron det n gi

Shortest Paths Problem. CS 362, Lecture 20. Today s Outline. Negative Weights

A DIVISIVE HIERARCHICAL CLUSTERING- BASED METHOD FOR INDEXING IMAGE INFORMATION

LinkGuide: Towards a Better Collection of Hyperlinks in a Website Homepage

Transcription:

Volume 3, Iue 11, November 2013 ISSN: 2277 128X International Journal of Advanced Reearch in Computer Science and Software Engineering Reearch Paper Available online at: www.ijarce.com Tak Aignment in Heterogeneou Environment with Advanced Scheduling Algorithm Mr. Minal Shahakar, Prof. Rajeh Bharati, Prof. Rupeh Mahajan Dept. of Computer Engg. DYPIET, Pune, India Abtract - Thi paper preent an innovative idea of cheduling the tak to their bet proceor to reduce the execution time of each tak by uing multiple heuritic technique. Thi paper preent hybrid heuritic technique which provide better olution of cheduling tak. Heuritic uch a MinMin+, MaxMin+ and Sufferage+ overcome the drawback of previouly ued heuritic method uch a MinMin, MaxMin and Sufferage a well a heuritic in thi paper provide better complexity a compare to previou heuritic method. Later the hybrid heuritic mean combination of different heuritic provide better performance without degrading the reult quality. Thi heuritic can alo be ued in heterogeneou environment more effectively to execute different et of tak on different proceor with different configuration. Key Term - MinMin, MaxMin, Sufferage, Tak Scheduling, Standard Deviation, Load Balancing. I. INTRODUCTION Scheduling i one of the important proce in the area of ditributed computing. It perform an important tep of mapping tak to different machine baed on the expected execution time. Normally an application i ued to define the piece of work of higher level in heterogeneou environment. Since thi application can generate everal number of job that can be divided into ubtak and provided to different proceor that hould get completed within minimum time o that the proceor ue can be made to aign different tak. Makepan i one of the mot important term in cae of mapping tak to their proceor uing different heuritic. Makepan i nothing but turnaround time that i maximum of completion time. An optimal chedule will be the one that minmize the makepan [1, 2]. The exiting heuritic provide the variou technique for aigning different tak to different proceor with minimum completion time. Thi exiting heuritic can be divided into two clae that are Online mode and Batch mode heuritic. In online mode, a tak i aigned to proceor a oon a it arrive at the cheduler [1]. Wherein Batch mode heuritic tak are not aigned to proceor immediately intead they are collected in to et of tak alo called a Metatak that are examined for aigning at precheduled time to different proceor alo called a mapping event. Since in thi work, batch mode i ued in very efficient way for mapping different independent tak to proceor. Alo thi exiting heuritic can be applied for heterogeneou environment effectively. Thi paper preent an efficient heuritic method that i hybrid method alo known a RASA algorithm. Overall the algorithm aim to minimize the idle time and makepan of tak. Thi paper alo involve the concept of Load Balancing [2, 6], wherein, once cheduling of tak i done uing ome heuritic the load balancing algorithm will take place to rechedule the tak to utilize all the reource in the heterogeneou environment [5]. Alo the propoed ytem in thi work contain variou heuritic method along with hybrid technology uch a Minmin+, Maxmin+, and Sufferage+. Each of thi heuritic provide better performance and alo increae complexity without degrading the olution quality. II. HEURISTICS DESCRIPTION Many algorithm have been deigned for mapping of tak on different proceor. Since exiting heuritic method aim at providing bet quality of olution without degrading ytem performance. At the ame time it aim to minimize the makepan, a it provide the tak aignment alternatively depending on expected execution time and tak aignment. In earlier ditributed computing ytem compried of homogeneou and dedicated reource. Thi heuritic will work well even for heterogeneou reource alo. Heterogeneou ytem provide with the facility of utilization of all available reource a load balancing concept that aim at keeping reource buy [5]. A. Heuritic Minimum Completion Time (MCT) aign tak to different proceor in arbitrary order, with minimum expected completion time for that tak. Since thi caue ome of the tak to be aigned to the proceor that do not have minimum execution time. For thi purpoe thi minimum completion time i defined in a way that combine the benefit of both opportunitic load balancing (OLB) and minimum execution time (MET) to provide better performance of tak mapping [3]. 2013, IJARCSSE All Right Reerved Page 1079

Minal et al., International Journal of Advanced Reearch in Computer Science and Software Engineering 3(11), Opportunitic Load Balancing (OLB) i pecially ued to keep all the proceor buy that i to make utilization of all the available reource by aigning tak in arbitrary order, to the next available proceor, without conidering tak expected execution time on that particular proceor but thi reult in poor makepan [6]. Minimum Execution Time (MET) aign tak to proceor in arbitrary order with bet expected execution time for that tak, without taking into conideration proceor availability. Since aigning the tak to it bet proceor provide better performance but caue evere load imbalancing and doe not provide upport for heterogeneou environment [4, 5]. B. Minmin heuritic Minmin heuritic tart with et of metatak that contain all unaigned tak that ha minimum completion time. Since for thi purpoe it require to calculate minimum completion time i.e. MCT of each tak in MT that i metatak. Once minimum completion time i found in the firt tep thi heuritic find minimum expected completion time of each tak in metatak and finally the tak with overall minimum expected completion time i found and aigned to the repective available reource [6]. When aigned to the reource it i removed from the metatak that i et of tak and thi proce i repeated until all the tak are aigned to reource. Since thi method provide eaiet way to aign tak to proceor but ha one drawback due to election of tak having minimum expected completion time, the tak with larget expected completion time remain unaigned for longer time and alo load i not balanced acro the ytem, due to which ome reource remain idle and thi alo reult in increae in makepan. C. Maxmin heuritic Maxmin heuritic i imilar to minmin heuritic wherein it alo tart with mapping independent tak to different machine. The et of tak i found that having minimum completion time from metatak. And the tak with overall maximum completion time alo known a makepan i elected and aigned to the available reource. Thi proce i continued until all the tak are aigned to repective available machine. Since thi heuritic provide the way of mapping tak to it bet machine with longer execution time firt allow thi tak to be executed concurrently with the tak that having horter execution time. Since thi mapping of tak to reource i better than minmin heuritic wherein the tak with maller execution time i elected and aigned to the available reource for execution and then tak with longer execution time are executed while everal machine it idle. Since maxmin heuritic provide better load balancing acro machine a well a better makepan [6, 7]. D. Sufferage Heuritic Sufferage heuritic differ with previou heuritic in the ene of tak election proce. Like minmin and maxmin it alo begin with et of unaigned tak that ha minimum completion time i.e. ufferage heuritic i alo baed on the concept of minimum completion time ince it differ from the previou heuritic in the ene it elect and aign the tak to the proceor on the bai of ufferage value and not minimum or maximum completion time. Since it compute econd MCT value intead of computing MCT value for each tak and calculate ufferage value which i defined a difference between MCT and econd MCT value of a tak i taken into account. Thi heuritic elect the tak with larget ufferage value and aign it to available reource. Thu ufferage heuritic differ from minmin and maxmin heuritic in the tak election policy [4, 5]. III. ADVANCE HEURISTICS Thi type of heuritic provide better performance than the previou heuritic without degrading olution quality. That i previouly mentioned heuritic increae time complexity by uing number of iteration for computing minimum completion time of each tak. Wherein, the advance heuritic thee MCT value of each tak are maintained eparately that reduce the number of iteration a well a reduce time complexity along with makepan and make proceing fater. A. Minmin+ heuritic Minmin+ heuritic ue different method for initialization of neceary variable and for election of tak with minimum completion time. Alo it maintain a eparate queue wherein all the MCT value are arranged in orted order. Thu minmin+ heuritic firt make neceary initialization, elect tak with minimum completion time for mapping to proceor and once tak i elected and aigned to proceor it i deleted from queue. For the implementing thi priority queue two alternative are being conidered that are binary heap and orted linear array, and alo ome operation are being ued like orting operation, deletion operation, and neceary check i made on the queue to know which tak ha not being yet aigned to available reource that i with minimum completion time. Hence the overall running time complexity i reduced. B. Maxmin+ heuritic Thi heuritic i imilar to minmin+ heuritic except it differ in the way of electing tak. Since it require MCT value of tak but aignment of tak to repective available reource i done on the bai of makepan that i maximum of completion time. Thu it doe firt election of tak to proceor aignment i done uing minmin+ heuritic only. That i neceary initialization of variable and tak election i done uing the ame method that are ued in minmin+ heuritic. The computed aignment i realized only if it doe not lead to increae in makepan of previou iteration. Since if computed aignment increae in makepan of previou iteration then tak aignment to proceor i recomputed 2013, IJARCSSE All Right Reerved Page 1080

Minal et al., International Journal of Advanced Reearch in Computer Science and Software Engineering 3(11), according to maxmin heuritic. Thi heuritic overcome drawback of maxmin heuritic of tak aignment problem to ame proceor by doing the combination of maxmin with minmin+ under a hybrid heuritic that i maxmin+. C. Sufferage+ heuritic Sufferage + working i imilar to ufferage heuritic ince to make applicable ufferage heuritic to large dataet it i combine with minmin+ heuritic under a new heuritic that i ufferage+. Here alo tak aignment i done according to minmin+ heuritic only that i initialization and tak election for aignment to proceor. Thi heuritic differ from previou heuritic in the ene that when aignment i computed, it i realized only if it doe not, lead to increae in makepan of previou iteration otherwie aignment i recomputed uing ufferage heuritic. D. Switcher Heuritic A the name indicate it i the combination of different heuritic alo known a hybrid heuritic. Switcher heuritic i baed on concept of tandard deviation value comparion with threhold value. Baed on thi it witche between heuritic that i if tandard deviation [6] value i le than threhold value than tak are conidered to be with minimum execution time and minmin heuritic i applied to aign the tak to available reource, otherwie maxmin heuritic i ued to aign the tak to available reource. Thi proce i repeated until all the tak are aigned to their repective available reource [3, 13]. There are many different type of hybrid algorithm that call alternatively different heuritic for mapping tak to their bet proceor. Thi type of heuritic alo maintain the proper load balance acro the proceor due to which all the available reource get fully utilized and no reource remain an idle. IV. STANDARD DEVIATION Standard deviation concept i pecially ued for hybrid heuritic that i combination of different heuritic. Wherein, the tandard deviation value i compared with threhold value to check which heuritic to be applied for mapping of tak to different reource [6]. Since the tandard deviation value i calculated on the bai of average of completion time of all tak, a mention below: avgct = Where avgct denote average of completion time that i um of all completion time of given tak and i nothing but index of tak. Uing thi average value tandard deviation i calculated a: i 1 d = (CT ij avgct) 2 Baed on above mentioned formulae tandard deviation i calculated. Since thi i compared with the threhold value in cae of hybrid heuritic wherein the multiple heuritic are called by algorithm alternatively for mapping tak to proceor. Thi hybrid heuritic will ue tandard deviation concept wherein if the calculated tandard deviation i le than threhold value then that particular tak i aigned uing the minmin heuritic to available reource otherwie the tak i aigned to available reource uing maxmin heuritic. Since after the aignment of tak to reource it will be deleted from et of tak that i metatak and the hybrid heuritic will repeat all the proce until all the tak are aigned to proceor. Thi tandard deviation mentioned above can alo be repreented in another way that i in the relation a mention below: i 1 CT ij Where E(x i ) denote the average of x i. d = E CT ij 2 E(CT ij ) 2 V. LOAD BALANCING Load Balancing concept take place in ditributed ytem to keep all the reource buy that i all the reource hould get utilized o that tak execution will become fater and time complexity will be reduced. Heuritic mentioned above like minmin and maxmin map different tak to different available reource efficiently but it doe not maintain proper load balancing among the reource due to which ome reource are utilized and ome remain idle. Thi load balancing concept can be applied to thi heuritic to get done execution fater [6]. Minmin heuritic elect tak with minimum completion time and allocate it to available reource, due to which tak with longer execution time remain unaigned although the reource i available that caue reource to remain idle. Similarly in maxmin heuritic tak with maximum completion time i elected and aigned to proceor, due to which maller tak are aigned after long time to available proceor [7]. Solution for above i to apply load balancing concept with thi type of heuritic. Thi can be done when tak are aigned to their reource that i once minmin heuritic i applied to aign tak to available reource uing minimum completion time, the load balancing method i applied again on thi aigned tak for recheduling it i.e. it may happen that minmin heuritic will ue ome reource to aign tak and ome remain idle then load balancing method will elect the tak with maximum completion time that will be le than makepan produced by Minmin heuritic and rechedule it to the reource that i available and not utilized yet o that execution of tak will be more fater [11]. Other tak maximum completion time i not le than makepan. So whichever tak ha maximum completion time le tan makepan i elected and recheduled to available reource. 2013, IJARCSSE All Right Reerved Page 1081

Minal et al., International Journal of Advanced Reearch in Computer Science and Software Engineering 3(11), VI. EXAMPLE OF LOAD BALANCING Conider a heterogeneou environment with two reource R 1 and R 2 and metatak that contain four different tak T 1, T 2, T 3 and T 4 a hown below in table1 that contain tak, reource along with expected execution time for mapping tak to their repective reource. Table 1: Reource and Tak with Expected Execution Time Tak Reource T 1 7 2 T 2 13 3 T 3 14 2 T 4 11 3 A hown in above table tak aignment i done to different proceor uing minmin heuritic i done in following way. 10 7 4 T 4 T 2 T 3 2 T 1 Figure 1: Mapping of Tak to Reource with Minmin Algorithm A hown in figure1 minmin heuritic will elect tak according to given execution time o tak T 1 will be aigned firt to Reource R 2, then tak T 3 will be aigned again on reource R 2, then tak T 2 will be aigned again on reource R 2 and finally the remaining tak that i tak T 4 will alo be aigned to reource R 2 only according to given expected execution time in Table1. Since on reource R 2 tak completion i fater than on reource R 1, o all the tak will be aigned on reource R 2 only. Once minmin heuritic i applied for mapping tak to available reource, load balancing technique i applied for again recheduling tak to make utilization of idle reource that minimize overall tak completion time that i makepan. 8 7 T 1 T 4 5 T 2 2 T 3 Figure 2: Recheduling of tak to Reource with Load Balancing method A hown in figure2 tak T 1 i recheduled to balance the load a it provide maximum completion time on reource R 1 a hown in Table1 a well a it i le than makepan produced by minmin heuritic. While remaining tak although have maximum completion time but are not le than makepan. So tak T 1 i recheduled on reource R 1 that reult in better makepan a compared to minmin heuritic. 2013, IJARCSSE All Right Reerved Page 1082

Minal et al., International Journal of Advanced Reearch in Computer Science and Software Engineering 3(11), VII. CONCLUSION The goal of thi paper wa to preent variou heuritic method like minmin, maxmin, ufferage, hybrid, load balancing technique in the field of ditributed ytem. The heuritic like minmin and maxmin are uitable for mall cale ditributed ytem but when number of tak increae than thee heuritic cannot chedule tak appropriately that affect on makepan which relatively become large. To overcome limitation of thee heuritic and make them applicable for large cale ditributed ytem, a new tak cheduling algorithm like minmin+, maxmin+ and ufferage+ along with hybrid heuritic are ued that alo maintain proper load balancing acro the ytem. Thi heuritic ue advantage of minmin and maxmin and cover there diadvantage. Thi tudy can be further extended by conidering tak heterogeneity and machine heterogeneity. REFERENCES [1] T. D. Braun,H. J. Siegel,N. Beck, L. L. Boloni, A comparion of eleven tatic heuritic for mapping a cla of independent tak onto heterogeneou ditributed computing ytem, J. Parallel Ditrib. Comput., vol. 61, no. 6, pp. 810837, 2001. [2] P. Luo, K. Lu, and Z. Shi, A reviit of fat greedy heuritic for mapping a cla of independent tak onto heterogeneou computing ytem, J. Parallel Ditrib. Comput., vol. 67, pp. 695714, 2007. [3] Kamali Gupta, Manpreet Singh, Heuritic Baed Tak Scheduling In Grid, International Journal of Engineering and Technology (IJET), vol. 4, pp. 254260, Aug-Sep 2012. [4] M. Mahewaran, S. Ali, H. J. Siegel, D. Hengen, and R. F. Freund, Dynamic mapping of a cla of independent tak onto heterogeneou computing ytem, J. Parallel Ditrib. Comput., vol. 59, pp. 107131, 1999. [5] H. J. Siegel and S. Ali, Technique for mapping tak to machine in heterogeneou computing ytem, J. Syt. Archit., vol. 46, no. 8, pp. 627639, 2000. [6] T. Kokilavani, Dr. D.I. George Amalarethinam, Load Balanced Min-Min Algorithm for Static Meta-Tak Scheduling in Grid Computing, International Journal of Computer Application, vol. 20, April 2011. [7] George Amalarethinam. D.I, Vaaheedha Kfatheen.S, Max-min Average Algorithm for Scheduling Tak in Grid Computing Sytem, International Journal of Computer Science and Information Technologie, Vol. 3, pp. 3659-3663, 2012. [8] K. Kaya, B. Ucar, and C. Aykanat, Heuritic for cheduling file-haring tak on heterogeneou ytem with ditributed repoitorie, J. Parallel Ditrib. Comput., vol. 67, no. 3, pp. 271285, 2007. [9] Doreen Hephzibah Miriam. D and Eawarakumar. K.S, A Double MinMin Algorithm for Tak Metacheduler on Hypercubic P2P Grid Sytem, IJCSI International Journal of Computer Science Iue, Vol. 7, Iue 4, No 5, July 2010. [10] He. X, X-He Sun, and Lazewki. G.V, QoS Guided Minmin Heuritic for Grid Tak Scheduling, Journal of Computer Science and Technology, Vol. 18, pp. 442-451, 2003. [11] Kamalam.G.K and Muralibhakaran.V, A New Heuritic Approach: MinMean Algorithm For Scheduling Meta- Tak On Heterogenou Computing Sytem, International Journal of Computer Science and Network Security, VOL.10 No.1, January 2010. [12] Sameer Singh Chauhan,R. Johi. C, QoS Guided Heuritic Algorithm for Grid Tak Scheduling, International Journal of Computer Application (09758887), pp 24-31, Volume 2, No.9, June 2010. [13] Singh. M and Suri. P.K, QPS A QoS Baed Predictive Max-Min, Min-Min, Switcher Algorithm for Job Scheduling in a Grid, Information Technology Journal, Vol. 7, pp. 1176-1181, 2008. [14] Yagoubi. B, and Slimani. Y, Tak Load Balancing Strategy for Grid Computing, Journal of Computer Science, Vol. 3, No. 3, pp. 186-194, 2007. 2013, IJARCSSE All Right Reerved Page 1083