Font Size: a A A

Performance Monitoring And Optimization Of A Virtualization Machine Xen

Posted on:2014-02-13Degree:MasterType:Thesis
Country:ChinaCandidate:J J PangFull Text:PDF
GTID:2248330395497154Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Nowadays, the distributed computing is developing all the time.The experts and researchers are taking great effort on the researchof the virtualization and the applications that based on thevirtualization. The basis of virtualization technology is virtualizedthe events or the objects so the computer virtualization technologyaims at how to abstract the computer resources. This method bring usmuch more convenience and it reduces the cost of engineeringapplications and scientific research. The virtualization technologyadd a Hypervisor software to the existing platform. To theapplications, this layer is transparent., different applications runits program on different virtual machines and they do not know thatthere are other virtual operating system is running. The transparencyassure that each virtual operating system is separated from eachother and the running process of each applications is independent.This paper’s work is based on the research of the Xen virtualmachine architecture and main technology, it proposed a newperformance optimization way and make a comparison of differentmonitoring tools. In the aspect of performance monitoring issue, weintroduce some common tools and construct the Xen platform to test Xenprof and Xencenter. From that work, the cpu state andvirtualization spending can be obtained.In the aspect of performance optimization, we concentrate on theXen VCPU scheduling algorithm and analysis its advantage andimperfact. According the different need, we proposed two improvedalgorithm: dynamic weighted aggregation Credit algorithm and mixedtype Credit algorithm which can be used in the SMP environment. Bothof the two algorithm has been tested in a simulation environment usedthe Schedsim toolkit.
Keywords/Search Tags:Virtualization, Xen, Performance Monitoring, Credit algorithm
PDF Full Text Request
Related items