Dynamic resource pool expansion and contraction in multiprocessing environments

• US 5,093,912 A
• Filed: 06/26/1989
• Issued: 03/03/1992
• Est. Priority Date: 06/26/1989
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Dong, Margaret A., Treiber, Richard K.
• Primary Examiner(s): Lee, Thomas C.
• Assistant Examiner(s): Treat, William M.
• Application Number:
  US07/371,563
• Time in Patent Office: 981 Days
• Field of Search: 364/200 MS File, 364/900 MS File
• US Class Current: 718/104
• CPC Class Codes: G06F 2209/5011 Pool
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources

Abstract

A method for expanding and contracting a resource pool dynamically, i.e. in response to demand for the resources, is disclosed. Resources are allocated for use from a single master pool 22. The overall pool of available resources is expanded by creating an extension pool and adding its resources to the master pool. The master pool can theoretically be expanded to the limits of the number of resources available in the system.

To contract the pool dynamically, first one of the extension pools is selected to be freed (FREE_-- POOL 66). Those resources owned by the freed pool but still in the master pool (i.e., not in use) are moved to the extension pool, preventing their reuse. Resources in use which are owned by the freed pool are labeled by changing their home address 52 so that they will be returned to the extension pool instead of to the master pool. Once all such resources have been returned to the freed extension pool it is deleted, thereby contracting the overall pool of resources.

Pseudocode implementations of the procedures for manipulating storage buffers and buffer pools are included.

Claims

1. A method for expanding and contracting a master pool of computer resources which are allocable for use in a computerized data processing system, comprising the steps executed by the system'"'"'s computer of:

(a) expanding the master pool by

(i) creating an extension pool containing at least one additional computer resource, and

(ii) adding the extension pool'"'"'s computer resources to the master pool; and

(b) contracting the master pool by

(i) transferring from the master pool to the extension pool any unallocated computer resource originating in the extension pool,

(ii) upon the deallocation of a computer resource originating in the extension pool, transferring such computer resource to the extension pool, and

(iii) deleting the extension pool when it contains all of the computer resources originating in it.

2. The method of claim 1, wherein the computer resources are buffers.

3. The method of claim 1, wherein at least one of the steps therein is intrinsically serialized using an atomic instruction.

4. The method of claim 3, wherein the atomic instruction is the IBM 370 Compare-Double-and-Swap instruction.

5. A method for managing the master pool of computer resources in a computerized data processing system, comprising the steps executed by the system'"'"'s computer of:

(a) creating an extension pool containing at least one additional computer resource;

(b) chaining the extension pool'"'"'s computer resources to the master resource pool;

(c) allocating (getting) the computer resources from the master pool for use by processes executing in the system;

(d) deallocating (freeing) the computer resources by the processes;

(e) returning said freed computer resources belonging to an active extension pool to the master pool;

(f) returning said freed computer resources belonging to a freed extension pool to said extension pool;

(g) transferring any unallocated computer resource belonging to a freed extension pool from the master pool to said extension pool; and

(h) deleting the freed extension pool when it contains all of the computer resources belonging to it.

6. The method of claim 5, wherein the steps of allocating and deallocating the computer resources are intrinsically serialized.

7. The method of claim 6, wherein the intrinsic serialization is provided by the use of an atomic instruction.

8. The method of claim 7, wherein the atomic instruction is the IBM 370 Compare-Double-and-Swap instruction.

9. The method of claim 5, wherein the master and extension pools are organized as chains of control elements.

10. The method of claim 9, wherein the computer resources are transferred between pools by heading their control elements to the destination pool'"'"'s chain, and deleting said control element from the originating pool'"'"'s chain.