Measuring and evaluating the performance of grid computing systems : throughput, response time and sytsem utilization

Abstract EN

Grid has emerged as popular paradigm distributed computing.

It enables aggregation of distributed resources for solving large-scale problems in science, engineering, and commerce.

In grid environments, the resources are usually geographically distributed in multiple administrative domains, managed and owned by different organizations with different policies, and interconnected by wide-area networks or the Internet.

The management of resources and scheduling of applications in such large-scale distributed systems is a complex undertaking.

In order to prove the effectiveness of resource brokers and associated scheduling algorithms, their performance needs to be evaluated under different scenarios such as varying number of resources and users with different requirements.

In a grid environment, it is difficult and even impossible to perform scheduler performance evaluation in a repeatable manner as resources and users are distributed across multiple organizations with their own policies.

To overcome this limitation, a toolkit called GridSim [2] has been developed a Java-based discrete-event grid simulation.

The toolkit supports modeling and simulation of heterogeneous grid resources (both time-shared and space-shared), users and application models.

It provides primitives for creation of application tasks, mapping of tasks to resources, and their management.

The main objective of this paper is to analyze some of the performance parameters such as the response time, the system utilization and throughput in the computational grid environment.

In our simulation, we conclude that higher the number of Gridlets (or higher the Gridlet length), longer the simulation time.

Simulation time of a number of Gridlets can be reduced with higher resource MIPS or higher number or resources.

System utilization is maximized with increasing the gridlet length.

Throughput is maximized with increasing number of gridlets.