Resource harvesting in scalable, fault tolerant, single system image clusters
AssigneeTandem Computers Incorporated
InventorsJames R. Alexander, Rory Foster, Robert R. Teisberg
Life of this patent
Estimating the term…
Loading patent…
AssigneeTandem Computers Incorporated
InventorsJames R. Alexander, Rory Foster, Robert R. Teisberg
Estimating the term…
A method is presented that enhances the survivability of system software components, even in the event of catastrophic failure of the computing element on which they reside. In particular, the combination of a distributed operating system (Non Stop Clusters) and a fault-tolerant interconnect (ServerNet) provides an environment conducive to posthumous recovery strategies that have been unavailable in previous distributed computing environments. The specific strategy outlined here is called resource harvesting, and involves a novel approach that retrieve critical data structures of memory from a failed computing element for reconstruction on a non-failed computing element, allowing such critical data structures to continue with their original function.
In a computing system including at least a first processor unit and a second processor unit, each having a memory element for storing data and each being communicatively interconnected to each of the other for communicating data, a method of harvesting predetermined portions of the data stored in the memory of the first processor unit, the method including the steps of:
In a computing system having a plurality of processor units each having a memory element for storing data and each being communicatively interconnected to each of the other processor units for communicating data, a method of harvesting predetermined portions of the data stored in the memory element of a one of the processor units, the method including the steps of:
Dependencies are inferred from the claim text.