160 likes | 269 Views
G-Monitor is a web-based portal designed for monitoring and steering application execution on global grids. Developed by Martin Placek and Rajkumar Buyya at the University of Melbourne's GRIDS Lab, it provides an easy-to-use interface for experiment management, job control, and resource monitoring. G-Monitor is scalable, capable of handling thousands of jobs and resources, and is adaptable for execution across different time zones. This system leverages technologies like Apache, Perl, and JavaScript, making grid computing accessible and efficient for users. Visit [gridbus.org](http://www.gridbus.org) to learn more.
E N D
WW Grid G-monitor: A Gridbus Web Portal for Monitoring and Steering Application Execution on Global Grids Martin Placek and Rajkumar Buyya Grid Computing and Distributed Systems (GRIDS) Lab.Dept. of Computer Science and Software EngineeringThe University of Melbournehttp://www.gridbus.org
G-Monitor • Introduction • Architecture • Implementation Notes • Gmonitor Snapshots • Conclusion
Introduction • Functionality/Requirements • Experiment Management and Control • Job control • Broker QoS • Experiment status • Job status • Resources status • Run from anywhere
Implementation • Technologies • Apache Web Server (CGI) • Perl • JavaScript • Nimrod-G Protocol • Issues • Timezones issue, as webclient, webserver and nimrod server can all be in different Timezones! (Fix:UTC) • Scalability - Handles large scale experiments (1000’s of Jobs,100’s of Resources.)
WW Grid G-monitor Usage: HPC Challenge Demo @ SC 2002 Australia North America GMonitor Melbourne+Monash U: VPAC, Physics ANL: SGI/Sun/SP2 NCSA: Cluster Wisc: PC/cluster NRC, Canada Many others Gridbus+Nimrod-G MEG Visualisation Solaris WS Internet @ SC 2002/Baltimore Europe Grid MarketDirectory ZIB: T3E/Onyx AEI: Onyx CNR: Cluster CUNI/CZ: Onyx Pozman: SGI/SP2 Vrije U: Cluster Cardiff: Sun E6500 Portsmouth: Linux PC Manchester: O3K Cambridge: SGI Many others Asia AIST, Japan: Solaris Cluster Osaka University: Cluster Doshia: Linux cluster Korea: Linux cluster
G-monitor - QoS • Deadline • Budget • Optimisation Algorithm • Start/Stop Experiment
G-monitor - Job Status & Resource Status • Job execution status • Further Attributes • Job Control • Resource Availability • Resource Cost (G$)
G-monitor - Experiment Status • Deadline (Assigned/Spent/Remaining) • Budget • Job Status Summary • Host Status Summary • Jobs Completed/Jobs Assigned
Conclusion • G-Monitor • Web based implementation • provides easy to use and ubiquitous interace to the Grid. • Easy to setup. • Scalable and places low overhead on the network • Try it and Download from http://www.gridbus.org/ • Future Work • Integrate G-monitor into a HTML framework (Eg. WebGUI) • Integrate further job/experiment statistics. • Integrate further job/experiment control.