Authors
Archana Nottamkandath
Publication date
2011/8
Institution
Vrije University
Description
In the cloud environment resource provisioning is driven from very coarse grained measurements. For instance, new resources are allocated to applications when the average load on the physical node exceeds a certain threshold. With the variety of services provided by the clouds, many companies and users are migrating their web applications to the cloud environment. As Web applications face significant variations in their request load, the amount of computational resources they use has to be dynamically adjusted. But the current resource provisioning algorithms do not take into account the actual response time of the deployed applications, and thus do not work well for the web applications. Thus we have solved the need for the high-precision monitoring of the web applications, by developing non-intrusive sensor modules and integrating with Ganglia monitoring framework for gathering high-precision metrics about the web applications, which aid in their efficient resource provisioning in the cloud environment. One of the main challenging metric which more clearly reflects the performance of the web application at the server side is the Client Perceived Response Time (CPRT). The results indicate that the Client Perceived Response Time measurements gathered at the server side are very similar to the actual response time as experienced at the client. The high-precision metrics have been integrated with the Ganglia monitoring framework with less overhead.
Scholar articles