Replicated control block handles for fault-tolerant computer systems

• US 6,704,884 B1
• Filed: 08/25/2000
• Issued: 03/09/2004
• Est. Priority Date: 02/16/2000
• Status: Expired due to Term

• Current Assignee: Metaswitch Networks Limited (Microsoft Corporation)
• Original Assignee: Adam Shepherd
• Inventors: Reekie, David William Maxwell, Shepherd, Adam Paul, Miller, Benjamin Mark Simon Kenneth, Tylee, Andrew John, Dancer, Colin Michael, Weeds, Nicolas Peter Everard, Brittain, Paul John
• Primary Examiner(s): Beausoliel, Robert
• Assistant Examiner(s): Bonzo, Bryce P.
• Application Number:
  US09/645,403
• Time in Patent Office: 1,292 Days
• Field of Search: 714/4, 714/6, 714/15, 714/18, 714/20
• US Class Current: 714/4.1
• CPC Class Codes: G06F 11/1675 Temporal synchronisation or re-synchronisation of redundant processing components
• CPC Class Codes: G06F 11/2023 Failover techniques
• CPC Class Codes: G06F 11/2048 where the redundant components share neither address space nor persistent storage
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups

Abstract

In a fault-tolerant system employing software fault-tolerance techniques, the identifiers used for control blocks or objects referenced on an interface between two co-operating partner software processes either require resynchronization after fail over of one or both of the partners or impose unwanted performance impacts on normal operation of the system. Replicated handles enhance the techniques currently used in such systems to avoid both the need for resynchronization and other potential performance impacts of the prior art.

Claims

1. A method for ensuring that the identifiers exchanged over an interface between partner software processes for the control blocks or objects used by said partner software processes have the same value in the primary copy and backup copies of said partner software processes executing on one or a plurality of processing units, said method comprising of the steps of

a. assigning said identifiers, known as handle values, from a plurality of handle number spaces associated with a process type of said partner process

b. allowing only said primary copy of said process types to assign said handle values from said handle number spaces

c. first means of replicating said handle value for each said control block or object when the internal state of said primary copy, including said control block or object, is replicated from said primary copy to one or a plurality of backup copies of said process type

d. coercing the handle value chosen by said backup copies of said process type to be the same as said handle value chosen by said primary copy

e. second means of replicating destruction of said handle value when said primary copy no longer requires said handle for said control block or object

f. resetting said handle number space used by a backup copy of said process type when said backup copy starts or after recovery from a previous failure of said replication of said handle values to said backup copy.

2. The method as set forth is claim 1, further comprising the step of associating a local identifier with said handle value, said step comprising the steps of

a. allowing said primary copy and said backup copies of said process type to register said local identifiers for said handle value, said local identifiers being independently assigned by said primary copy and each of said backup copies and stored in memory or other storage media

b. allowing said primary copy and said backup copies to retrieve said local identifiers from said memory using said handle value to query the database of said local identifiers.

3. The method as set forth in claim 2, with a further step of ensuring that said handle value is not immediately reused after said handle value is destroyed by said primary copy.

4. The method as set forth in claim 3, wherein immediate reuse is prevented by means of forming part of said handle value from a use count, said use count being updated each time the other parts of said handle value are reused.

5. The method as set forth in claim 3, wherein immediate reuse is prevented by means of assigning said handle values from said handle number space on a least-recently-used basis.

6. The method as set forth in claim 5, wherein immediate reuse is further prevented by means of forming part of said handle value from a use count, said use count being updated each time the other parts of said handle value are reused.

7. The method as set forth in claim 3, wherein uniqueness of said handle value is guaranteed by a further step of assigning said handle number spaces to a unique said process type from within a group comprising a plurality of said process types such that said handle values are guaranteed to be unique across said group.

8. The method as set forth in claim 7, wherein said first means and said second means are achieved by message exchanges between said primary copy and said backup copies.

9. The method as set forth in claim 7, wherein said first means and said second means are achieved by use of shared memory or other shared storage media accessible to said primary copy and said backup copies.

10. The method as set forth in claim 2, with a further step of assigning said handle number spaces to a unique said process type from within a group comprising a plurality of said process types such that said handle values are guaranteed to be unique across said group.

11. The method as set forth in claim 2, wherein said first means and said second means are achieved by message exchanges between said primary copy and said backup copies.

12. The method as set forth in claim 2, wherein said first means and said second means are achieved by use of shared memory or other shared storage media accessible to said primary copy and said backup copies.