Lock manager on disk

1. A method of managing resources in a database management system, comprising:

receiving, at a first master node of a plurality of master nodes, a first lock request from a second node, wherein the first lock request is a request for a lock on a first shared resource;

wherein each master node of said plurality of master nodes has shared access to a plurality of shared resources;

wherein each master node of said plurality of master nodes masters a respective subset of said plurality of shared resources;

wherein said first shared resource belongs to a first subset of said plurality of shared resources that is mastered by said first master node;

said first master node granting the lock to said second node;

while said second node holds said lock, said second node causing said first master node to modify said first shared resource; and

wherein the method is performed by one or more computing devices.

2. The method of claim 1 , further comprising:

receiving, at said first master node, a second lock request from a third node, wherein the second lock request is a request for a lock on a second shared resource;

receiving a first modify request from said second node, wherein the first modify request is a request to modify said first shared resource;

subsequent to receiving the first modify request, receiving a second modify request from said third node, wherein the second modify request is a request to modify said second shared resource; and

modifying said second shared resource before modifying said first shared resource.

3. The method of claim 2 , wherein each modify request comprises an expected completion time.

4. The method of claim 2 , further comprising:

based at least in part on a prioritization scheme, determining whether the first modify request or the second modify request should be honored first.

5. The method of claim 4 , wherein the prioritization scheme is based on one or more of: a) an expected completion time associated with the request; b) priority attributes associated with the request's type; or c) a network or system state.

6. The method of claim 1 , further comprising:

receiving a second modify request from said second node, wherein the second modify request is a request to modify said first shared resource; and

subsequent to receiving the second modify request, receiving a first recall request from said second node, wherein the first recall request causes said first master node to refrain from modifying a second shared resource.

7. The method of claim 1 , further comprising:

caching said first shared resource at said first master node;

receiving a second modify request from said second node, wherein the second modify request is a request to modify said first shared resource; and

modifying the cached copy of said first shared resource.

8. The method of claim 1 , further comprising:

said first master node assuming mastership over said first shared resource, wherein said first shared resource was previously managed by a third master node of said plurality of master nodes.

9. The method of claim 8 , wherein the step of said first master node assuming mastership over said first shared resource is performed in response to detecting a state of the third master node.

10. The method of claim 1 , further comprising:

the second node generating an in-memory copy of said first shared resource.

11. A non-transitory computer-readable storage medium storing instructions which, when executed, cause one or more processors to perform the steps of:

receiving, at a first master node of a plurality of master nodes, a first lock request from a second node, wherein the first lock request is a request for a lock on a first shared resource;

wherein each master node of said plurality of master nodes has shared access to a plurality of shared resources;

wherein each master node of said plurality of master nodes masters a respective subset of said plurality of shared resources;

wherein said first shared resource belongs to a first subset of said plurality of shared resources that is mastered by said first master node;

said first master node granting the lock to said second node; and

while said second node holds said lock, said second node causing said first master node to modify said first shared resource.

12. The computer-readable storage medium of claim 11 , wherein the instructions further include instructions for:

receiving, at said first master node, a second lock request from a third node, wherein the second lock request is a request for a lock on a second shared resource;

receiving a first modify request from said second node, wherein the first modify request is a request to modify said first shared resource;

subsequent to receiving the first modify request, receiving a second modify request from said third node, wherein the second modify request is a request to modify said second shared resource; and

modifying said second shared resource before modifying said first shared resource.

13. The computer-readable storage medium of claim 11 , wherein the instructions further include instructions for:

based at least in part on a prioritization scheme, determining which modify request should be honored first.

14. The computer-readable storage medium of claim 12 , wherein the instructions further include instructions for:

based at least in part on a prioritization scheme, determining whether the first modify request or the second modify request should be honored first.

15. The computer-readable storage medium of claim 14 , wherein the prioritization scheme is based on one or more of: a) an expected completion time associated with a request; b) priority attributes associated with the request's type; or c) a network or system state.

16. The computer-readable storage medium of claim 11 , wherein the instructions further include instructions for:

receiving a second modify request from said second node, wherein the second modify request is a request to modify said first shared resource; and

subsequent to receiving the second modify request, receiving a first recall request from said second node, wherein the first recall request causes said first master node to refrain from modifying a second shared resource.

17. The computer-readable storage medium of claim 11 , wherein the instructions further include instructions for:

caching said first shared resource at said first master node;

receiving a second modify request from said second node, wherein the second modify request is a request to modify said first shared resource; and

modifying the cached copy of said first shared resource.

18. The computer-readable storage medium of claim 11 , wherein the instruction further include instructions for:

said first master node assuming mastership over said first shared resource, wherein said first shared resource was previously managed by a third master node of said plurality of master node.

19. The computer-readable storage medium of claim 18 , wherein the step of said first master node assuming mastership over said first shared resource is performed in response to detecting a state of the third master node.

20. The non-transitory computer-readable storage medium of claim 11 , further storing instructions which, when executed, cause the one or more processors to perform the steps of:

the second node generating an in-memory copy of said first shared resource.