Sei sulla pagina 1di 21

Grid and Cluster

Performance analysis of Grid and Cluster Computing based on running benchmark circuit
Guided By: Dr.K.Najeeb Group Members: Absheer KK Nikhil AT Shafeeque MKM Subhash PM Varghese K Renny
Government Engineering College Sreekrishnapuram Palakkad

Grid and Cluster

Overview

1 2 3 4 5 6 7 8

Problem denition Motivation Introduction Types of computing Tools used Work done Conclusion Future work

Grid and Cluster

Problem denition

Setup Beowulf cluster Setup Grid computing environment using Globus toolkit Analyse performance using standard benchmarks

Grid and Cluster

Motivation
Millions of computers are being found useless throughout the world Heterogeneous computers over the whole world can be used to provide a lot of CPU power and data storage capacity Cheap and easy way to take o-the-shelf components and combine them into a single supercomputer

Grid and Cluster

Introduction

Types of Computing
Parallel Computing Distributed Computing Cluster Computing Grid Computing Utility Computing Cloud Computing Pervasive computing

Grid and Cluster

Contd..

Cluster Computing
A Cluster is a type of parallel or distributed processing system, which consists of a collection of interconnected stand-alone computers cooperatively working together as a single integrated computing resource The components of a Cluster are connected to each other through fast local area networks Cluster Middleware High performance and availability

Grid and Cluster

Figure: 32 node Linux Cluster

Grid and Cluster

Contd..

Grid Computing
Grid computing is a term referring to the combination of computer resources from multiple administrative domains to reach a common goal. Many networked loosely coupled computers acting together to form a super virtual computer. Middleware divides and apportion pieces of a program among several computers

Grid and Cluster

Figure: Grid Computing structure

Grid and Cluster

Contd..

Comparison between Grid and Cluster


Cluster
Uses aordable building blocks Typically homogeneous conguration

Grid
More dynamic and usually heterogeneous system Distributed control

Grid and Cluster

Tools and Platform used

Tools
Message Passing Interface(MPI). Parallel Virtual Machine(PVM). Globus tool kit. NAS Parallel Benchmark.

Platform
Linux

Grid and Cluster

Contd..

Message Passing Interface(MPI)


Language independent communications protocol used to program parallel computers Allows processes to communicate with one another by sending and receiving messages Provides source code portability and support heterogenous computing

Grid and Cluster

Contd..

Parallel Virtual Machine(PVM)


Software tool for parallel networking of computers Allow a network of heterogeneous Unix and/or Windows machines to be used as a single distributed parallel processor PVM consists of a run-time environment and library for message-passing, task and resource management, and fault notication

Grid and Cluster

Contd.. Globus
Open source toolkit for building computing grids developed and provided by the Globus Alliance. Includes software services and libraries for resource monitoring, discovery, and management, plus security and le management Globus Toolkit is an implementation of the following standards
Open Grid Services Architecture (OGSA) Open Grid Services Infrastructure (OGSI) Web Services Resource Framework (WSRF)

Latest stable release is Globus 5.0.2.

Grid and Cluster

Contd..

NAS Parallel Benchmark


Developed and maintained by NASA Advanced Super Computing(NAS) Division Used for the performance evaluation of highly parallel super computer Includes following benchmark circuits
MultiGrid(MG) Conjugate Gradient(CG) Fast Fourier Transform(FT) Block Tridiagonal(BT)

Grid and Cluster

Work done

Studied about dierent types of computing Constructed Beowulf cluster using two nodes Studied about Parallel Programming Successfully executed MPI Parallel program in Beowulf Cluster Studied about NAS Parallel Benchmark

Grid and Cluster

Contd..

Beowulf Architecture
Multi-computer architecture which can be used for parallel computations Consists of one server node, and one or more client nodes connected together via Ethernet or some other network Server node controls the whole cluster and serves les to the client nodes Uses GNU/Linux or Solaris operating system, Parallel Virtual Machine (PVM) and Message Passing Interface (MPI)

Grid and Cluster

Figure: Beowulf Cluster architecture

Grid and Cluster

Conclusion

Both Grid and Cluster allows ecient use of idle resources Highly ecient and dont have single points of failure Lot of the major institutions have developed grid farms using blade servers Grid Computing will drive future internet growth

Grid and Cluster

Future work

Integrate Beowulf Cluster using more nodes Implement Grid Computing Environment using Globus tool kit Run benchmarks in both Grid and Cluster Evaluate the performance

Grid and Cluster

Reference

M Irving, G Taylor and P Hobson Plug in to Grid Computing , Power and Energy Magazine, IEEE, Volume 2, Issue 2, Mar-Apr 2004, pp40-44 Hsun-Chang Chang,Kuan-Ching Li and Yaw-Ling Lin Performance issues of grid computing based on dierent architecture cluster computing platforms,19th international conference on Advanced information Networking and Applications 2005. Thomas Sterling Beowulf Cluster Computing with Linux http://www.globus.org, Visited on 20-12-2010

Potrebbero piacerti anche