Dr. D. Ravindran, M. Ramya

Similar documents
n Explore virtualization concepts n Become familiar with cloud concepts

Enhancing Cloud Computing Scheduling based on Queuing Models

Multi-Threading. Hyper-, Multi-, and Simultaneous Thread Execution

A Hierarchical Load Balanced Fault tolerant Grid Scheduling Algorithm with User Satisfaction

DISTRIBUTED ALGORITHM FOR MULTI-AGENT ENVIRONMENT

Analysis of Server Resource Consumption of Meteorological Satellite Application System Based on Contour Curve

An Improved Shuffled Frog-Leaping Algorithm for Knapsack Problem

Switching Hardware. Spring 2018 CS 438 Staff, University of Illinois 1

UNIVERSITY OF MORATUWA

Prevention of Black Hole Attack in Mobile Ad-hoc Networks using MN-ID Broadcasting

Load Balancing in Cloud Computing : A Survey

What are Information Systems?

Security of Bluetooth: An overview of Bluetooth Security

REVIEW AND COMPARISON OF TASKS SCHEDULING IN CLOUD COMPUTING

The Value of Peering

A Study on the Performance of Cholesky-Factorization using MPI

Reliable Transmission. Spring 2018 CS 438 Staff - University of Illinois 1

Task scenarios Outline. Scenarios in Knowledge Extraction. Proposed Framework for Scenario to Design Diagram Transformation

Software development of components for complex signal analysis on the example of adaptive recursive estimation methods.

HADOOP: A NEW APPROACH FOR DOCUMENT CLUSTERING

CORD Test Project in Okinawa Open Laboratory

Energy Efficinet Cluster Head Selection and Data Convening in Wireless Sensor Networks

UH-MEM: Utility-Based Hybrid Memory Management. Yang Li, Saugata Ghose, Jongmoo Choi, Jin Sun, Hui Wang, Onur Mutlu

1&1 Next Level Hosting

Ones Assignment Method for Solving Traveling Salesman Problem

Outline n Introduction n Background o Distributed DBMS Architecture

Evaluation scheme for Tracking in AMI

Keywords Software Architecture, Object-oriented metrics, Reliability, Reusability, Coupling evaluator, Cohesion, efficiency

Transitioning to BGP

Data diverse software fault tolerance techniques

Morgan Kaufmann Publishers 26 February, COMPUTER ORGANIZATION AND DESIGN The Hardware/Software Interface. Chapter 5

Data Protection: Your Choice Is Simple PARTNER LOGO

n Learn how resiliency strategies reduce risk n Discover automation strategies to reduce risk

3D Model Retrieval Method Based on Sample Prediction

Politecnico di Milano Advanced Network Technologies Laboratory. Internet of Things. Projects

CMSC Computer Architecture Lecture 11: More Caches. Prof. Yanjing Li University of Chicago

Chapter 4 Threads. Operating Systems: Internals and Design Principles. Ninth Edition By William Stallings

Optimization for framework design of new product introduction management system Ma Ying, Wu Hongcui

Firewall and IDS. TELE3119: Week8

Pruning and Summarizing the Discovered Time Series Association Rules from Mechanical Sensor Data Qing YANG1,a,*, Shao-Yu WANG1,b, Ting-Ting ZHANG2,c

A QoS Provisioning mechanism of Real-time Wireless USB Transfers for Smart HDTV Multimedia Services

Sectio 4, a prototype project of settig field weight with AHP method is developed ad the experimetal results are aalyzed. Fially, we coclude our work

A SOFTWARE MODEL FOR THE MULTILAYER PERCEPTRON

FREQUENCY ESTIMATION OF INTERNET PACKET STREAMS WITH LIMITED SPACE: UPPER AND LOWER BOUNDS

SCI Reflective Memory

Copyright 2016 Ramez Elmasri and Shamkant B. Navathe

DCMIX: Generating Mixed Workloads for the Cloud Data Center

Adaptive Resource Allocation for Electric Environmental Pollution through the Control Network

GPUMP: a Multiple-Precision Integer Library for GPUs

arxiv: v2 [cs.ds] 24 Mar 2018

UNIVERSITY OF MORATUWA

DESIGN AND ANALYSIS OF LDPC DECODERS FOR SOFTWARE DEFINED RADIO

COMPUTER ORGANIZATION AND DESIGN The Hardware/Software Interface. Chapter 4. The Processor. Part A Datapath Design

Introduction to OSPF. ISP Training Workshops

Quorum Based Data Replication in Grid Environment

COSC 1P03. Ch 7 Recursion. Introduction to Data Structures 8.1

Lecture Notes 6 Introduction to algorithm analysis CSS 501 Data Structures and Object-Oriented Programming

Optimization of Priority based CPU Scheduling Algorithms to Minimize Starvation of Processes using an Efficiency Factor

Operating System Concepts. Operating System Concepts

A Modified Multiband U Shaped and Microcontroller Shaped Fractal Antenna

COMPUTER ORGANIZATION AND DESIGN The Hardware/Software Interface. Chapter 4. The Processor. Single-Cycle Disadvantages & Advantages

Copyright 2016 Ramez Elmasri and Shamkant B. Navathe

Service Oriented Enterprise Architecture and Service Oriented Enterprise

S. Mehta and K.S. Kwak. UWB Wireless Communications Research Center, Inha University Incheon, , Korea

Enhancing Efficiency of Software Fault Tolerance Techniques in Satellite Motion System

Ontology-based Decision Support System with Analytic Hierarchy Process for Tour Package Selection

Realistic Storage of Pending Requests in Content-Centric Network Routers

Session Initiated Protocol (SIP) and Message-based Load Balancing (MBLB)

Announcements. Reading. Project #4 is on the web. Homework #1. Midterm #2. Chapter 4 ( ) Note policy about project #3 missing components

Improvement of the Orthogonal Code Convolution Capabilities Using FPGA Implementation

Next generation IP- based multimedia services on cable TV networks

Data Warehousing. Paper

Lecture 28: Data Link Layer

Adaptive Graph Partitioning Wireless Protocol S. L. Ng 1, P. M. Geethakumari 1, S. Zhou 2, and W. J. Dewar 1 1

Evaluation of Distributed and Replicated HLR for Location Management in PCS Network

Basic allocator mechanisms The course that gives CMU its Zip! Memory Management II: Dynamic Storage Allocation Mar 6, 2000.

Image Segmentation EEE 508

IMP: Superposer Integrated Morphometrics Package Superposition Tool

Goals of the Lecture UML Implementation Diagrams

Appendix D. Controller Implementation

GE FUNDAMENTALS OF COMPUTING AND PROGRAMMING UNIT III

Harris Corner Detection Algorithm at Sub-pixel Level and Its Application Yuanfeng Han a, Peijiang Chen b * and Tian Meng c

. Written in factored form it is easy to see that the roots are 2, 2, i,

performance to the performance they can experience when they use the services from a xed location.

1. SWITCHING FUNDAMENTALS

Redundancy Allocation for Series Parallel Systems with Multiple Constraints and Sensitivity Analysis

Morgan Kaufmann Publishers 26 February, COMPUTER ORGANIZATION AND DESIGN The Hardware/Software Interface. Chapter 5.

How do we evaluate algorithms?

Extending The Sleuth Kit and its Underlying Model for Pooled Storage File System Forensic Analysis

VMbuddies: Coordinating Live Migration of Multi-Tier Applications in Cloud Environments

Python Programming: An Introduction to Computer Science

Secure Software Upload in an Intelligent Vehicle via Wireless Communication Links

On Nonblocking Folded-Clos Networks in Computer Communication Environments

Detection and Elimination of Fake Access Points in WLAN using Multi Agents Sourcing Methodology

FCM-SEP (Fuzzy C-Means Based Stable Election Protocol) for Energy Efficiency in Wireless Sensor Networks

Euclidean Distance Based Feature Selection for Fault Detection Prediction Model in Semiconductor Manufacturing Process

Algorithms for Disk Covering Problems with the Most Points

Outline. CSCI 4730 Operating Systems. Questions. What is an Operating System? Computer System Layers. Computer System Layers

Architectural styles for software systems The client-server style

Stone Images Retrieval Based on Color Histogram

Transcription:

2016 IJSRSET Volume 2 Issue 4 Prit ISSN: 2395-1990 Olie ISSN : 2394-4099 Themed Sectio: Egieerig ad Techology Moitorig ad Normalizatio of the Loaded Servers By Load Balacig Dr. D. Ravidra, M. Ramya School of Computig Sciece, St. Joseph s College, Trichy, Tamil Nadu Idia ABSTRACT Cloud computig is a ew computatioal model which is primarily based o distributed computig. The services ad applicatios are accessible to the differet cliets usig proper iteret protocol suit ad etworkig stadards. Though it has may advatages the cliet has to cosider also the drawbacks of the services provided by the cloud. Some of the issues have to cosidered like security, scalability, cost, SLA, etc. Most of the cliet users have to otice the performace of the cloud that the provider is issuig to the cliet. This paper describes about oe of the factors whe cosidered for the performace of the services provided. Scalability is the commo factor which affects the satisfactio of the cloud users. By reviewig the cloud scalability factors, moitorig the server that are distributed over the etwork ad its load balacig issue of the servers. This paper greatly deals with the ormalizatio of the cloud servers. Keywords : Performace; Scalability; Load Balacig; Moitorig; Normalize; Status I. INTRODUCTION Cloud computig are outlied as a computig surroudigs wherever computig wats by oe party are ofte outsourced to a differet party ad oce would like be arise to use the computig power or resources like iformatio or emails, they will access them via web[1]. Cloud computig is a model for eablig uiversal, suitable, o demad etwork accesses to a shared pool of cofigurable computig resources like etworks, servers, storage, applicatios, ad services that ca be rapidly equipped ad released with less maagemet effort or service provider itervetio. Moitorig of Cloud is a task of paramout importace for both Providers ad Cosumers. O the oe side, it is a key tool for cotrollig ad maagig hardware ad software ifrastructures; o the other side, it provides iformatio ad Key Performace Idicators (KPIs) for both platforms ad applicatios. Moitorig is clearly istrumetal for all the activities covered by the role of Cloud Auditor. I more geeral terms, Cloud Computig ivolves may activities for which moitorig are a essetial task. II. METHODS AND MATERIAL A. Related Work M. Kriushath et al.(2015)[2] describes the auto scalig values ad that settig dyamic threshold values i a cloud eviromet should utilize the available resources completely ad prevets the physical server damage. It maipulates the provider to accommodate more users i a physical server ad also reduces the cost of the service. I this paper, the authors elaborate their cocept i the area to set a dyamic threshold value for the physical server, load balacer behavior idetifier mechaism is proposed to geerate the rule ad provide the resources dyamically. Major drawback is that it ca t be used i cloud data. Abhijit Aditya et al. (2015)[3] presets the basics of cloud computig like it s characteristics, deploymets models, service models. They are describig the each service delivery models characteristics, its vedor types their advatages ad disadvatages. The they describig about where the load problems are occurrig i the system ad so the challeges i keepig mid. The it describes each ad every types IJSRSET1624171 Received : 19 August 2016 Accepted : 25 August 2016 July-August 2016 [(2)4: 781-785] 781

of algorithm i load balacig separately. Their properties, advatages, disadvatages are also described. They specially described about these algorithms based o the time factor. Po-Huei Liag et al.(2015)[4] presets a framework for global server usig for load balacig of the web sites i a cloud with two-level load balacig model. The proposed framework is iteded for adjustig a ope-source load-balacig system ad while the customers eed more load balacers for icreasig the availability, this framework allows the etwork service provider to deploy the load balacer i differet data ceters dyamically. Further they described the load balacig algorithms with the various cloud service providers alog with its commuicatio iterface. C. Load Balacig Approaches Static ad dyamic are the two type of load balacig approaches used i cloud computig[7,8]. Static Approach This approach is maily defied i the desig or implemetatio of system. Static load balacig algorithm divides the traffic equivaletly amog all users. It uses oly iformatio about the average behavior of the system. These are much simpler ad igore the curret state or the load of the ode i the system. Dyamic Approach Radha Ramai (2015)[5] presets a cocept of Cloud Computig alog with load balacig. The mai thig is cosidered i this paper is load balacig algorithm. There are various metioed algorithms i cloud computig which cosists of may factors like scalability, ehaced resource utilizatio, high performace ad improved respose time. Further this paper provides the isight about the policies, characteristics, goals, curret state classificatio, eed for load balacig. They have proposed a frame work for givig the ew algorithm. B. Scalability Scalability is the capability of a system, etwork, or process to hadle a growig amout of work, or its potetial to be elarged i order to accommodate that growth[6]. For example, it ca refer to the capability of a system to icrease its total output uder a icreased load. Cloud balacig is a computer etworkig method to distribute work load across multiple computer or a computer cluster, etwork lik, cetral processig uits, disk drivers or other resources to achieve optimal resource utilizatio, maximize throughput, miimize respose time ad avoid overload. Load balacig helps i prevetig bottleeck of system. First step, algorithm is desiged based o performace i heterogeeous eviromet of hosts The ext process is to study the above algorithm with the effect of CPU utilizatio. The based o the algorithm result, the algorithm will make the decisio about the ormalizig factor. I this approach, the curret state of the system was cosidered durig load balacig decisio. It is more suitable for widely distributed system such a cloud computig Dyamic approach has two parts[7] Cetralized Approach: Oly a sigle ode is resposible for maagig ad distributio withi the whole system. Distributed Approach: Each ode idepedetly builds its ow load vector. Vector collectig load iformatio of other ode. All decisio is made locally usig local load vector. D. Factors To Be Cosidered Performace metrics also play a major role i moitorig techiques for resource maagemet. As per the survey, it is cosidered as secod issue i cloud computig. Poor performace ca be caused by lack of resources such as disk space, limited badwidth, lower CPU speed, memory, etwork coectios etc. The data itesive applicatios are more challegig to provide proper resources. There is a series of factors that affect the performace such as [9]: Security. Recovery ad Fault tolerace. Service level agreemets. Badwidth. Storage capacity. Physical memory. Disk capacity. Processor Power. 782

Availability. Number of users ad Workload. Usability. Scalability. Locatio, data ceters ad their distace from a user s locatio. Ad there is a series of criteria for evaluatig the performace such as[10]: Average respose time per uit time. Average waitig time per uit time. Workload to be serviced per secod (Mbps) or a uit of time. Throughput (Req / Sec). The average time of processig (exe / sec). Percetage of CPU utilizatio. The umber of requests executed per uit time. The umber of requests per uit time buffer. The umber of rejected requests per uit time. E. Workload Calculatio Workload coditio i load balacer varies from small to heavy request[11]. As per the performace details, the proposed algorithm is based o the percetage of CPU utilizatio. To predict the workload ad allocate the suitable resources, load balacig ad scalig mechaisms are used. Scalig ca be doe i two ways called reactive ad proactive respectively. Reactive techiques are always time cosumig also mislay user satisfactio[12]. Proactive techiques are always preferable to avoid such complicatios. Workload has bee predicted usig CPU usage, Memory usage ad Network usage. CU = CPU utilizatio V = Total CPU s used Memory Load Memory is aother importat part of the computer all the work by the cliet is stored ad retrieved o ad from the memory respectively. So without the memory usage the workig process of the whole system become dazed. Memory load is measured as where MEM USED = memory used TOT MEM = total memory Memory = MEM USED/TOT MEM Network Load Network is the techique by which the systems are itercoected. Network systems ca be loaded if there are too may systems i a coditio to execute. The cliet has to cosider the etwork badwidth whe the cliet is goig to calculate the load. Network load measured as Network = NET BANDWIDTH / TOT BANDWIDTH where NET BANDWIDTH = etwork badwidth used TOT BANDWIDTH = total amout of etwork badwidth. CPU Load: F. Dyamic Normalized Algorithm The system will have may processes to execute i a sigle system. The task maager will show may variables like the CPU utilizatio of the resources, memory usage, etc. The CPU load plays the major role i the executio of etire system. Its utilizatio has great impact o the performace of a system. The two distiguished algorithm has proposed for the cloud service providers. The system is maitaied o ormalized mode o these two proposals [13]. The CPU usage has got by the performace couter. From the each IP address the CPU load has bee calculated by the above algorithm[14,15]. CPU load is measured as where = the umber of odes, CPU = (CU/V) Aget1: Begi While all the servers are i ruig state Get CPU usage from each system 783

Calculate the CPU load The CPU load has bee aalyzed for the differet request If (CPU load>=upper threshold) Status is OVERLOAD Else if (CPU load <Upper threshold && CPU load>ormal value) Status is NORMAL Else Status is UNDERFLOW Ed Aget 2: If (Status==OVERFLOW) Server has to Sleep for a while Else Server works i balaced mode III. RESULTS AND DISCUSSION A. Process Diagram To make the system as i the ormalized state, a check has to be performed[16]. The process of makig the system as ormalized is give i a flow diagram as follows Figure 2. Chart for CPU load IV. CONCLUSION Load balacig is cosidered as the mai factor i performace issues. Here, load balacig is aalysed o the IaaS level. There may be differet load factors the user ca cosider. I this paper, the CPU load ad the way to ormalize the executio of the CPU are measured. The whole implemetatio improves the efficiecy of the distributed CPU s. Though it has good efficiecy, the request has to wait for the overloaded server to become load balaced. Therefore, the task legth ad waitig time is icreased. Here this cocept is implemeted oly o CPU load; i future, the authors ca try this algorithm usig all the workloads of the system. V. REFERENCES Figure 1. Process diagram B. Aalysis of The Load Balacer The resources are aalyzed i this paper ad it is based o oly oe factor i.e. CPU performace [16]. The followig performace chart is studied based upo the time ad the CPU load. The load has bee differetiated based o the above algorithm. This chart shows the CPU utilizatio of the various servers coected i the etwork. [1] Mohd Hairy Mohamaddiah, Azizol Abdullah, Shamala Subramaiam, Masida Hussi, A Survey O Resource Allocatio Ad Moitorig I Cloud Computig, Iteratioal Joural Of Machie Learig Ad Computig, Vol4, No1, February 2014, DOI: 10.7763/IJMLC.2014.V4.382 [2] M.Kriushath, DrLArockiam, Load Balacer Behavior Idetifier (Lobbi) For Dyamic Threshold Based Auto-Scalig I Cloud, ICCCI -2015, Ja08 10, 2015, Coimbatore, INDIA. [3] Abhijit Aditya, Uddalak Chatterjee, Sehasis Gupta, A Comparative Study Of Differet Static Ad Dyamic Load Balacig Algorithm I Cloud Computig With Special Emphasis O Time Factor, Iteratioal Joural Of Curret Egieerig Ad Techology E-ISSN 2277 4106, P-ISSN 2347 5161 2015. [4] Po-Huei Liag1 Ad Jia-Mi Yag, Evaluatio Of Two-Level Global Load 784

Balacig Framework I Cloud Eviromet, Iteratioal Joural Of Computer Sciece & Iformatio Techology (IJCSIT) Vol 7, No 2, April 2015, DOI:10.5121/Ijcsit.2015.7201 1. [5] Radha Ramai Malladi, A Approach to Load Balacig I Cloud Computig, Iteratioal Joural of Iovative Research i Sciece, Egieerig ad Techology, ISSN(Olie): 2319 8753, Vol4, Issue 5, May 2015. [6] Priyaka PKukade, Geetajali Kale, Survey Of Load Balacig Ad Scalig Approaches I Cloud, Iteratioal Joural Of Emergig Treds & Techology I Computer Sciece (IJETTCS), Volume 4, Issue 1, Jauary-February 2015 ISSN 2278-6856 [7] Weiku Wag, Giuliao Casale, Evaluatig Weighted Roud Robi Load Balacig For Cloud Web Services. [8] Jiteder Grover, Shivagi Katiyar, Aget Based Dyamic Load Balacig I Cloud Computig, https://www.researchgate.et/publicatio/271544 765 [9] Aarti Sigh, Dimple Jueja, Maisha Malhotra, Autoomous Aget Based Load Balacig Algorithm I Cloud Computig, ICACTA 2015, Procedia Computer Sciece 45 ( 2015 ) 832 841 [10] Hardi SSaghavi, DrTejas PPatalia, Load Balacig Algorithms For The Cloud Computig Eviromet: A Review, Joural Of Iformatio, Kowledge Ad Research I Computer Egieerig, ISSN: 0975 6760 Nov 14 To Oct 15 Volume 03, Issue 02 Page 591. [11] Amal Zaouch, Faouzia Beabbou, Load Balacig For Improved Quality Of Service I The Cloud, Iteratioal Joural Of Advaced Computer Sciece Ad Applicatios, Vol6, No7, 2015 [12] Shikha Gupta, Suma Saghwa, Load Balacig I Cloud Computig: A Review, Iteratioal Joural Of Sciece, Egieerig Ad Techology Research (IJSETR), Volume 4, Issue 6, Jue 2015, [13] MsShilpa D.More, ProfArti Mohapurkar, Load Balacig Strategy Based O Cloud Partitioig Cocept, Multidiscipliary Joural Of Research I Egieerig Ad Techology, Volume 2, Issue 2, Pg.424-431, ISSN: 2348-6953 [14] Smita Salukhe, SSSaakki, Load Rebalacig For Distributed File Systems I Clouds, IJAFRC, Volume 2, Special Issue - NCRTIT 2015, Jauary 2015ISSN 2348 4853295 2015 [15] Nadeem Shah, Mohammed Farik, Static Load Balacig Algorithms I Cloud Computig: Challeges & Solutios, Iteratioal Joural Of Scietific & Techology Research Volume 4, Issue 10, October 2015 ISSN: 2277-8616. [16] Akshada Bhujbal, Prajakta Jakate, Maasi Wagh, Madhura Pise, Prof.M.V.Marathe, Load Balacig Model i Cloud Computig, Iteratioal Joural of Emergig Egieerig Research ad Techology Volume 3, Issue 2, February 2015, PP 1-6 ISSN 2349-4395 (Prit) & ISSN 2349-4409 (Olie). 785