Expert system for computer system resource management

• US 5,367,473 A
• Filed: 11/26/1991
• Issued: 11/22/1994
• Est. Priority Date: 06/18/1990
• Status: Expired due to Term

• Current Assignee: George Washington University
• Original Assignee: Bell Communications Research, Inc. (Telefonaktiebolaget LM Ericsson)
• Inventors: Eichelman, II, Fredric R., Huleatt, Thomas O., Irgon, Adam E., Langion, Martha S., Chu, Lynn H., Dragoni, Jr., Anthony H.
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US07/798,496
• Time in Patent Office: 1,092 Days
• Field of Search: 364/550, 364/551.05, 395/908, 395/909, 395/916, 395/917, 395/575, 395/700
• US Class Current: 702/186
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: G06F 11/34 Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
• CPC Class Codes: Y10S 706/916 Electronic or computer, internal or network, circuit

Abstract

An expert system methodology for operating a computer system is disclosed to facilitate tuning, that is, resource balancing of the various resources comprising a large-scale software system. The computer system is adapted to store a knowledge base derived from experts who manually tune the large-scale system. The system also stores site-specific information and specifications on components configuring the large-scale system. Performance measurements collected from the large-scale system on a time consistent basis are merged, utilizing the knowledge base, with reference data derived from the site-specific information and the specifications. As a result of the merger, the status of the large-scale system is classified as to its operability. If the status is in a predefined class requiring tuning, the computer system is arranged to generate programming code which will effectuate the required changes in the resources to thereby reconfigure the large-scale system.

Claims

1. A method in a data processing system for improving the responsiveness of a dedicated, application-specific host computer being sized to insure a concurrent transaction backlog during peak hour, the host computer including disks, disk control units, and files on disks controlled by the disk control units, and having measurable central processing unit (CPU) and input/output (I/O) service times, the method comprising the steps of

defining a set of operating principles relating to the operation of host system by capturing operational and descriptive information supplied by a host system expert who acts as a mentor,

transforming said set of operating principles into an executable knowledge base and storing said knowledge base in a memory of an expert system support process having memory,

measuring the CPU and I/O service times of the host computer under control of said support process storing the service times in said support process memory,

processing the service times to convert the service times to system state data organized into a state space of processing rates and queuing times, and

applying said executable knowledge base to said system state data to obtain host throughput-response time trade-off adjustment information so as to modify the transaction concurrency of the host computer as determined by said trade-off adjustment information.

2. The method as recited in claim 1 further comprising the steps of

loading into said expert system support process the current I/O configuration arrangement of the host system, and

sequencing through said set of operating principles embedded in said executable knowledge base with the I/O service times and said I/O configuration arrangement to obtain I/O adjustment information so as to re-allocate the current allocation of files to disks as determined by said I/O adjustment information.

3. The method as recited in claim 2 wherein said step of processing the service times includes the steps of processing the service times utilizing a self-calibrating central server model, solving the model, and producing said system state data.

4. The method as recited in claim 2 wherein said step of sequencing through said set of operation principles with the I/O service times includes the step of generating instructions to reconfigure the host computer so as to change the current allocation of files to disks.

5. A method in a data processing system for improving the responsiveness of a dedicated, application-specific host computer being sized to insure a concurrent transaction backlog during peak hour, the host computer including disks, disk control units, and files on the disks controlled by the disk control units, and having measurable central processing unit (CPU) and input/output (I/O) service times, the method comprising the steps of

defining a set of operating principles relating to the operation of the host computer by capturing operation and descriptive information supplied by a system expert who acts as a mentor,

transforming said set of operating principles into an executable knowledge base and storing said knowledge base in an expert system support process having memory,

measuring the CPU and I/O service times of the host computer under control of said support computer and storing the service times in said support process,

processing the service times to convert the service times to system state data organized into a state space of processing rates and queueing times,

applying said executable knowledge base to said system state data to obtain host throughput-response time trade-off adjustment information so as to modify the transaction concurrency of the host computer as determined by said trade-off adjustment information,

loading into said expert system support process the current I/O configuration arrangement of the host computer, and

sequencing through said set of operational principles embedded within said executable knowledge base by said transforming step with the I/O service times and said I/O configuration arrangement to obtain I/O adjustment information so as to re-allocate the current allocation of files to disks as determined by said I/O adjustment information.

6. The method as recited in claim 5 wherein said step of processing the service times includes the steps of processing the service times utilizing a self-calibrating central server model, solving the model, and producing said system state data.

7. The method as recited in claim 5 wherein said step of sequencing through said set of operational principles with the I/O service times includes the step of generating instructions to change the host computer so as to re-allocate the current allocation of files to disks.

8. A data processing system for improving the responsiveness of a dedicated, application-specific host computer being sized to insure concurrent transaction backlog during peak hour, the host computer including disks, disk control units, and having measurable central processing unit (CPU) and input/output (I/O) service times, the processing system comprising

an expert system support process having a memory, said support process being connected to the host computer,

means for capturing operational and descriptive information in said expert system process supplied by a host system expert who acts as a mentor so as to define a set of operating principles relating to the operation of the host system,

means for transforming said set of operating principles into an executable knowledge base and storing said knowledge base in said memory of said expert system support process,

means for measuring the CPU and I/O service times of the host computer under control of said expert system support process and storing the service times in said expert system support process,

means for processing the service times to convert the service times to system state data organized into a state space of processing rates and queueing times and

means for applying said executable knowledge base to said system state data to obtain host throughput-response time trade-off adjustment information so as to modify the transaction concurrency of the host computer as determined by said trade-off adjustment information.

9. The data processing system as recited in claim 8 further comprising

means for loading into said support process the current I/O configuration arrangement of the host system, and

means for sequencing through said set of operational principles embedded in said executable knowledge base with the I/O service times and said I/O configuration arrangement to obtain I/O adjustment information so as to re-allocate the current allocation of files to disks as determined by said I/O information.

10. The data processing system as recited in claim 9 wherein said means for processing the service times includes means for processing the service times utilizing a self-calibrating central server model, solving the model, and producing said system state data.

11. The data processing system as recited in claim 9 wherein said means for sequencing through said set of operation principles with the I/O service times includes means for generating instructions to reconfigure the host computer so as to change the current allocation of files to disks.

12. A data processing system for improving the responsiveness of a dedicated, application-specific host computer being sized to insure a concurrent transaction backlog during peak hour, the host computer including disks, disk control units, and files on the disks controlled by the disk control units, and having measurable central processing unit (CPU) and input/output (I/O) service times, the processing comprising

an expert system support process having a memory, said support process being connected to the host computer,

means for capturing of operational and descriptive information supplied by a system expert who acts as a mentor so as to define a set of operating principles relating to the operation of the host system,

means for transforming said set operating principles into an executable knowledge base and storing said knowledge base in said memory of said expert system support process,

means for measuring the CPU and I/O service times of the host computer under the control of said expert system support process and storing the service times in said expert system support process,

means for processing the service times to convert the service times to system state data organized into a state space of processing rates and queueing times,

means for applying said executable knowledge base to said system state data to obtain host throughput-response time trade-off adjustment information so as to modify the transaction concurrency of the host computer as determined by said trade-off adjustment information,

means for loading into said expert support process the current I/O configuration arrangement of the host computer, and

means for sequencing through said set of operational principles embedded within said executable knowledge base with the I/O service times and said I/O configuration arrangement to obtain I/O modification information so as to re-allocate the current allocation of files to disks as determined by said modification information.

13. The data processing system as recited in claim 12 wherein said means for processing the service times includes means for processing the service times utilizing a self-calibrating central server model, solving the model, and producing said system state data.

14. The data processing system as recited in claim 12 wherein said means for sequencing through said set of operation principles with the I/O service times includes means for generating instructions to reconfigure the host computer so as to change the current allocation of files to disks.