Failure resistant multiple computer system and method

A failure resistant multiple computer system in which a plurality of computers each has a corresponding independent local memory which is unable to be accessed by any other of said computers, each of said computers simultaneously operates a corresponding portion of an application program written to be executed only on a single computer, and each is connected via a communications network to permit updating of corresponding memory locations, said system including:

• a local memory independence means associated with each said local memory to satisfy all read requests from the local memory of the computer making the read request,
• a reachability means to categorize memory locations of said local memories into a first category in which the local memory locations are replicated as to contents stored in the memory locations in selected ones, or all, of said computers and therefore require updating via said communications network with changes to corresponding memory locations of other computers, to maintain substantial memory coherence, and into a second category in which the local memory locations are not replicated and are present only in the local computer and therefore no updating is required, and wherein said system further includes:
• a failure detection means connected to each said computer to detect failure of any one of said multiple computers, and
• a reachability modifier connected to said failure detection means and to said reachability means to modify said reachability means by modifying said first category to remove therefrom, if present, any reference by the failed computer whereby no attempt is made to update any first category memory locations of said failed computer,
• wherein the local memory independence means is unable to be accessed by other computers and all read requests by the computer making the read request are satisfied locally from the local independent memory.

A failure resistant method of operating a plurality of computers, the method comprising:

• configuring each of said plurality of computers with their own corresponding independent local memory so that each independent local memory is unable to be accessed by any other ones of said plurality of computers and all read requests from each said computer are satisfied from the local memory of the computer making the read request;
• operating on each of said computers substantially simultaneously a corresponding portion of an application program written to execute on only a single computer;
• connecting each of said plurality of computers via a communications network to permit updating of corresponding memory locations in the local memory of each of the plurality of computers;
• categorizing the memory locations of said local memories into:
• (a) a first reachability category in which the local memory locations are replicated as to contents stored in the memory locations in selected ones, or all, of said computers and therefore require updating via said communications network with changes to corresponding memory locations of the other computers to maintain substantial memory coherence;
• and
• (b) a second category in which the local memory locations are not replicated and are present only in the local computer and therefore no updating is required;
• detecting failure of any one of said plurality of computers;
• and
• modifying said first reachability category to remove therefrom, if present, any reference to the failed computer, and making no attempt to update any first reachability category memory locations of said failed computer.

A failure resistant multiple computer system comprising:

• a plurality of computers each having a corresponding independent local memory which is unable to be accessed by any other of said computers, and each of said computers configured to simultaneously operate a corresponding different portion of an application program written to be executed only on a single computer;
• each said computer being connected to each other computer via a communications network to permit updating of corresponding memory locations;
• a local memory independence means associated with each said local memory to satisfy all read requests from the local memory of the computer making the read request,
• a reachability means to categorize memory locations of said local memories into:
• (a) a first category in which the local memory locations are replicated as to contents stored in the memory locations in selected ones, or all, of said computers and therefore require updating via said communications network with changes to corresponding memory locations of other computers, to maintain substantial memory coherence;
• and
• (b) a second category in which the local memory locations are not replicated and are present only in the local computer and therefore no updating is required;
• a failure detection means connected to each said computer to detect failure of any one of said plurality of computers, and
• a reachability modifier connected to said failure detection means and to said reachability means to modify said reachability means by modifying said first category to remove therefrom, if present, any reference by the failed computer and making no attempt to update any first reachability category memory locations of said failed computer,
• wherein the local memory is unable to be accessed by other computers and all read requests by the computer making the read request are satisfied locally from the local independent memory.