An apparatus and method shares predictive failure information between computer system in a computer network. The shared predictive failure information allows dynamically rerouting a network request to avoid a computer system that may fail according to its predictive failure information. According to a first embodiment, if the requested resource on the network has predictive failure information that indicates the resource may soon fail, a message is returned to the requesting computer with information that includes possible alternative sites from which the information may be obtained. If there is an alternative site, the requesting computer system may access the alternative site, thereby avoiding the computer system that may soon fail. If there is no alternative site, the requesting computer system may return an error message, or may simply access the original resource on the chance that is has not yet failed. According to a second embodiment, a router in the network may indicate one or more alternative paths to a resource if the predictive failure information for the router indicates it may soon fail. The requesting computer system may then access the requested resource via the alternative path. In this manner, predictive failure information can be used in rerouting network traffic between computer systems on a network to minimize the effect of a failing computer system.
This patent application is divisional of my earlier patent application of the same title, Ser. No. 09/580,048, filed on May 26, 2000, now U.S. Pat. No. 6,802,020 which is a continuation-in-part of my earlier patent application entitled "APPARATUS AND METHOD FOR SHARING PREDICTIVE FAILURE INFORMATION ON A COMPUTER NETWORK", Ser. No. 09/522,127, filed on Mar. 9, 2000 now U.S. Pat. No. 6,609,212. Both of these earlier patent applications are incorporated herein by reference.
A large population of mass-produced devices (80) such as a particular model of computer hard disk drive, are distributed around the world. Each device (80) includes an arrangement for collecting failure analysis data of the device (50). Each device (80) is arranged to transmit this data to the device manufacturers server (10) via the internet (20). The server (10) analyses the data in order to determine trends in failure performance of the population of devices in order to improve future designs and to provide updated software for distribution to the devices (80) via the internet (20).
