Modified machine architecture with machine redundancy

• US 20070100828A1
• Filed: 10/18/2006
• Published: 05/03/2007
• Est. Priority Date: 10/25/2005
• Status: Abandoned Application

• Current Assignee: Waratek Pty Limited
• Original Assignee: Waratek Pty Limited
• Inventors: Holt, John
• Application Number:
  US11/583,962
• Publication Number:
  US 20070100828A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/0709 in a distributed system consisting of a plurality of standalone computer nodes, e.g. clusters, client-server systems
• CPC Class Codes: G06F 11/073 in a memory management context, e.g. virtual memory or cache management memory management G06F12/00; testing of static memory units G11C29/00
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms

Abstract

A multiple computer system in which a single application program, written to execute on only a single computer, runs on multiple computers is disclosed. Each computer (M1, . . . Mn) has a substantially identical local memory structure. Synchronizing locks are used to ensure that only one computer is able to write to a local memory location and all other computers are prohibited to writing to their corresponding memory location. In the event of failure of a computer holding such a lock, the lock is arranged to be released. The released lock can then be re-allocated to another computer which has not failed. In this way failure of one, or a sequence of, computers, does not result in failure of the whole computer system.

Claims

1. In a multiple computer environment in which an application program written to execute only on a single computer runs simultaneously on a plurality of computers each of which has a local memory in which globally named objects, assets or resources are locally substantially replicated and in which a synchronizing lock corresponding to the global name is acquired and released in sequence by any computer utilizing one of said objects, assets or resources, said lock authorizing the acquiring computer to update the local contents of the locked object, asset or resource and preventing all other computers accessing their corresponding local object, asset or resource, the improvement comprising the step of:

following computer failure of any computer which has acquired but not released any specific lock,

(i) releasing said specific lock, whereby said application program running conducted by the non-failed ones of said computers can continue by allocation of said lock to a non-failed one of said computers in due course, if necessary.

19. A computer program product comprising a set of program instructions stored in a storage medium and operable to permit a plurality of computers to carry out the method up as claimed in claim 1 or form the computer system as claimed in claim 7.

20. A plurality of computers interconnected via a communications network and operable to ensure carrying out of the method as claimed in claim 1.

21. An application program stored in a computer readable storage medium and modified to carry out the method as claimed in claim 1.

2. The improved method as claimed in claim I wherein at the time of said computer failure at least one other computer is awaiting allocation of said specific lock, said method comprising the further step of:

(ii) allocating said specific lock to the, or one of, said computer(s) awaiting allocation.

3. The improved method as claimed in claim 2 comprising the further steps of:

(iii) detecting said computer failure,

(iv) determining whether the failed computer held any unreleased locks,

(v) determining the identity of said other computer(s) awaiting allocation, and

(vi) allocating said specific lock to the computer identified in step (v).

4. The improved method as claimed in claim 3 comprising the further steps of:

(vii) maintaining a table of allocated locks which table includes the identity of the machine to which each lock has been allocated,

(viii) carrying out step (v) by consulting said table, and

(ix) updating said table after carrying out step (vi).

5. The improved method as claimed in claim 2 comprising the further steps of:

(x) requiring any lock requesting computer to interrogate each possible lock holding computer as to whether it holds the specific lock, and

(xi) utilizing the absence of a positive answer within a predetermined time to trigger allocation of said specific lock to said requesting computer whereby any lock holding computer which fails is incapable of said positive answer and thereby permits allocation of said lock to said lock requesting computer.

6. The improved method as claimed in claim 5 comprising the further step of:

(xii) requiring said lock requesting computer to repeat step (x) after a predetermined delay following receipt of a positive answer to step (x).

7. A multiple computer system in which an application program written to execute only on a single computer runs simultaneously on said multiple computers each of which has a local memory in which globally named objects, assets or resources are locally substantially replicated and in which a synchronizing lock corresponding to the global name is acquired and released in sequence by any computer utilizing one of said objects, assets or resources, said lock authorizing the acquiring computer to update the local contents of the locked object, asset or resource and preventing all other computers accessing their corresponding local object, asset or resource, wherein said system includes a computer failure detector to detect failure of any one of said computers and release means to release any lock acquired but not released by a failed computer, whereby said application program running conducted by the non-failed ones of said computers can continue allocation of said lock to a non-failed one of said computers in due course, if necessary.

8. The system as claimed in claim 7 and including re-allocation means operable in the event of said detection of failure of one of said computers, and at that time of detection there being at least one other computer awaiting allocation of said lock, to re-allocate said lock to the, or one of, said computers awaiting allocation.

9. The system as claimed in claim 8 including identification means to identify any unreleased lock(s) held by a failed computer and to identify any computers awaiting allocation of said unreleased lock(s).

10. The system as claimed in claim 9 wherein said identification means comprises a table of locks allocated and a queue of computers awaiting lock allocation.

11. The system as claimed in claim 7 wherein said failure detection means interrogates each possible lock holding computer and absence of a reply to said interrogation is deemed to constitute computer failure.

12. The system as claimed in claim 11 wherein said detection means repeats said interrogation at predetermined intervals in response to a reply thereto indicating no failure of the interrogated computer.

13. A single computer intended to operate in a multiple computer system which comprises a plurality of computers each having a local memory and each being interconnected via a communications network wherein different portions of at least one application program each written to execute on only a single computer, each execute substantially simultaneously on a corresponding one of said plurality of computers, and at least one memory location is replicated in the local memory of each said computer, said system further comprising updating means associated with each said computer to in due course update each said memory location via said communications network after each occasion at which each said memory location has its content written to, or re-written, with a new content,

said single computer comprising:

a local memory having at least one memory location intended to be updated via a communications port connectable to said communications network,

updating means to in due course update the memory locations of other substantially similar computers via said communications port;

lock means associated with said local memory to acquire a lock on an object, asset or resource of said local memory, a computer failure detector to detect failure of another computer, and

release means to release any lock acquired but not released by a failed computer, whereby said application program portion executing on said single computer can acquire said lock of failed computers in due course, if necessary.

14. The system as claimed in claim 13 including re-allocation means to re-allocate said acquired lock to another computer awaiting allocation.

15. The system as claimed in claim 14 including identification means to identify any unreleased lock(s) held by a failed computer and to identify any computers awaiting allocation of said unreleased lock(s).

16. The system as claimed in claim 15 wherein said identification means comprises a table of locks allocated and a queue of computers awaiting lock allocation.

17. The system as claimed in claims 13 wherein said failure detection means sends an interrogation to said communications port and absence of a reply to said interrogation is deemed to constitute computer failure.

18. The system as claimed in claim 17 wherein said detection means repeats said interrogation at predetermined intervals in response to a reply thereto.

22. In a single computer capable of interoperating with at least one other computer coupled to said single computer at least intermittently via a communications network to form a multiple computer system having a plurality of computers wherein each computer has a local memory, a method for handling a lock of an object, asset, or resource comprising:

executing at least a portion of at least one application program written to execute on only a single computer and modified to execute substantially simultaneously on one of said plurality of computers;

replicating at least one memory location in the local memory of each of said plurality of computers;

updating each said memory location of said other computer in due course via said communications network after each occasion at which a memory location has a memory content written to, or re-written, with a new content;

acquiring a lock on an object, asset or resource of said local memory;

detecting a failure of another computer, and

releasing any lock acquired but not released by a failed computer, so that said application program portion executing on said single computer can acquire said lock of failed computers in due course.

23. A method as in claim 22, further comprising performing the modification of the at least a portion of the at least one application program written to execute on only a single computer to execute substantially simultaneously on one of said plurality of computers.

25. A computer program as in claim 23, further comprising instructions which, when executed on the computer, perform a method for handling a lock of an object, asset, or resource, said method further comprising the step of:

performing the modification of the at least a portion of the at least one application program written to execute on only a single computer to execute substantially simultaneously on one of said plurality of computers.

24. A computer program recorded on a memory device comprising instructions which, when executed on a computer, perform in at least one single computer capable of interoperating with at least one other computer coupled to said single computer at least intermittently via a communications network to form a multiple computer system having a plurality of computers wherein each computer has a local memory, a method for handling a lock of an object, asset, or resource, said method comprising the steps of:

replicating at least one memory location in the local memory of each of said plurality of computers;

updating each said memory location of said other computer in due course via said communications network after each occasion at which a memory location has a memory content written to, or re-written, with a new content;

acquiring a lock on an object, asset or resource of said local memory;

detecting a failure of another computer, and

releasing any lock acquired but not released by a failed computer, so that said application program portion executing on said single computer can acquire said lock of failed computers in due course.