Redistribution of operating environments for the redeployment of grid clients

摘要

A client-server network environment is temporarily transformed into a grid computing environment through the deployment of a network-booted grid operating system to each of a plurality of network clients. Each participating client operates in the client-server mode under a local operating system. At an appropriate time, a server transmits to each client a thin-client boot image (a dynamically built customized image or an image obtained from a store of predefined boot images) along with a grid application for which the thin-client has been optimized for peak performance. Each client re-boots using the thin-client image, thereby entering the grid mode and allowing substantially all of its computing resources to be dedicated to executing the grid application. At a later time, the server directs that each client terminate the grid application and re-boot in its respective local operating system, reverting again to the client-server mode.

主权项

1. A method for temporarily transforming a client-server network into a grid network, comprising:
allowing each of a plurality of client devices to operate under a local operating system in a client-server mode; at a first predetermined time, configuring a server, with which the client devices are coupled through a network, to operate under a thin-client operating system, whereby the server becomes a master node on the network; for each client device:
determining whether a boot image is available for deployment to the client device;after determining that a boot image is not available for the client device, generating a boot image for the client device;directing that the client device initiate a network boot in a thin-client mode, whereby the client device becomes a processing node for the server;transmitting a thin-client image and a grid application to the processing node, whereupon the client device boots the thin-client image; andlaunching the grid application on the processing node, whereby the processing node operates in a grid mode with substantially all resources of the server and memory of the processing node being available for use by the grid application, the grid application able to launch different grid applications on different processing nodes; and at a second, later predetermined time, directing that each processing node re-boot into its respective local operating system.