Performance Monitoring Of Optica] Network For Grid Applications

Photonic grid is the combinative product of grid technology and optical networks. Inaddition to including the huge distributed heterogeneous resources of traditional grid, p hotonicgrid also have many optical network resources, such as optical links, optical nodes and so on.Due to the dynamic changes of the photonic grid resources, it is very important to monitor theresources of photonic grid in order to provide clients muc h better effective and flexible services.The state about the load of photonic grid resources can be acknowledged by resource monitoring,and then the scheduling of photonic grid resources will be optimized, and the load of both grid’snodes and network will be balanced well. The potential problems and bottlenecks of the networkas well as equipment could be found easily, and then the photonic grid system will be maintainedand upgraded.MDS (Monitoring and the Discovery System) is one of components in the existing gridtoolkit——Globus Toolkit, which uses an extensible framework to monitor the static anddynamic information of various resources in the grid environment. However, the monitorperformance of MDS is not satisfactory and its monitoring data query services provided to theuser are limited. Meanwhile, MDS is short of the optical network resources, especially thenetwork bandwidth. In order to make the Globus Toolkit adapt to the photonic grid monitoring, asecondary development should be developed further.This article mainly discussed the photonic grid monitoring service. The monitor function ofMDS in the Globus Toolkit can be optimized by utilizing Ganglia and NWS. The realizedperformance through optimized monitoring system includes: real-time monitoring of the thephotonic grid nodes’s processor, memory, hard disk utilization, I/O load, network traffic andother status messages; utilizing NWS’s functions which include real-time data measurement andforecasting for the optical network bandwidth and network delay, the optical network resourcescan become a kind resources, like other grid resources released to GRAM, and then GRAM canbetter optimize resource scheduling; providing clients with a web monitoring data curve view.The data monitored by photonic grid is stored and transmitted in XML format, so it could beadaptable to the Globus Toolkit heterogeneous characteristics.