Rules-based method of and system for optimizing server hardware capacity and performance

• US 6,557,035 B1
• Filed: 03/30/1999
• Issued: 04/29/2003
• Est. Priority Date: 03/30/1999
• Status: Expired due to Term

• Current Assignee: Lenovo International Limited (Lenovo Group Ltd.)
• Original Assignee: International Business Machines Corporation
• Inventors: McKnight, Gregory Joseph
• Primary Examiner(s): Barot, Bharat
• Application Number:
  US09/281,383
• Time in Patent Office: 1,491 Days
• Field of Search: 709/202-203, 709/217-219, 709/223-224, 709/231-235, 709/105, 710/52, 710/56-57, 710/60
• US Class Current: 709/224
• CPC Class Codes: G06F 11/3409 for performance assessment
• CPC Class Codes: G06F 11/3433 for load management allocation of a server based on load conditions G06F9/505; load rebalancing G06F9/5083; redistributing the load in a network by a load balancer H04L67/1029
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis
• CPC Class Codes: G06F 2201/81 Threshold

Abstract

A method of optimizing server hardware performance and predicting server hardware bottlenecks monitors server hardware utilization parameters over a selected time period and computes the averages of the measurements. The method then compares the computed averages to thresholds. If some of the computed averages are equal to or greater than the threshold, the method reports a performance bottleneck and provides a recommended solution for the bottleneck. The method predicts a future server hardware performance bottleneck by computing running averages of the measured server utilization parameters. The method uses a linear regression analysis to determine a trend in the running averages and compares the trend to threshold values to predict the occurrence of a performance bottleneck.

Claims

1. A method of optimizing server performance, which comprises the computer implemented steps of:

monitoring at least one server hardware utilization parameter over a selected time period;

computing an average of said at least one server utilization parameter over said selected time period;

comparing said average to a threshold value for said at least one server utilization parameter;

reporting a performance bottleneck if said average is greater than said threshold value, wherein said reporting comprises sending an output to an output device of said server that is accessible to a system administrator;

recommending a solution to said performance bottleneck; and

outputting a recommendation to the output device of said server.

2. The method as claimed in claim 1, wherein said utilization parameter is CPU utilization.

3. The method as claimed in claim 1, wherein said utilization parameter is memory utilization.

4. The method as claimed in claim 1, wherein said utilization parameter is disk utilization or disk queue depth.

5. The method as claimed in claim 1, wherein said utilization parameter is LAN byte throughput.

6. The method as claimed in claim 1, wherein said utilization parameter is LAN packet throughput.

7. The method as claimed in claim 1, including the computer implemented steps of:

monitoring said at least one utilization parameter over a plurality of selected time periods;

computing the average of said at least one utilization parameter for each selected time period;

determining a trend over time for said average;

predicting a performance bottleneck utilizing said trend; and

outputting said prediction to said output device.

8. The method as claimed in claim 7, wherein said step of determining said trend includes the computer implemented step of applying linear regression to said average for each of said selected time period.

9. The method as claimed in claim 7, wherein said selected time periods are overlapping, whereby said averages are running averages.

10. The method of claim 1, wherein said server has a plurality of drives, said method further comprises:

checking which one of said plurality of drives has a lowest running average utilization; and

recommending that some of the work load be migrated to the drive with the lowest running average utilization.

11. A method of optimizing server performance, which comprises the computer implemented steps of:

monitoring each of a plurality of server hardware utilization parameters over a selected time period;

computing an average of each of said plurality of server utilization parameters over said selected time period;

dynamically comparing each average to a threshold value for a corresponding utilization parameter;

reporting a performance bottleneck if at least one of said average is greater than the threshold value for the corresponding utilization parameter, wherein said reporting comprises sending an output to an output device of said server that is accessible to a system administrator;

recommending a solution to said performance bottleneck; and

outputting a recommendation to the output device of said server.

12. The method as claimed in claim 11, wherein said utilization parameters are selected from a plurality of measurable utilization parameters including CPU utilization, memory utilization, disk queue depth, LAN byte throughput, and LAN packet throughput, said method including the computer implemented step of:

applying a set of rules to said average values to report performance bottlenecks.

13. The method as claimed in claim 12, wherein said step of applying a set of rules includes the computer implemented step of:

reporting a CPU bottleneck if;

said average CPU utilization is equal to or greater than a CPU utilization threshold;

said average memory utilization is less than a memory utilization threshold;

said average disk queue depth is less than a disk queue depth threshold;

said LAN byte throughput is less than a LAN byte throughput threshold; and

,

said LAN packet throughput is less than a LAN packet throughput threshold.

14. The method as claimed in claim 12, wherein said step of applying a set of rules includes the computer implemented step of:

reporting a memory bottleneck if;

said average CPU utilization is less than a CPU utilization threshold;

said average memory utilization is equal to or greater than a memory utilization threshold;

said average disk queue depth is less than a disk queue depth threshold;

said LAN byte throughput is less than a LAN byte throughput threshold; and

,

said LAN packet throughput is less than a LAN packet throughput threshold.

15. The method as claimed in claim 12, wherein said step of applying a set of rules includes the computer implemented step of:

reporting a disk bottleneck if;

said average CPU utilization is less than a CPU utilization threshold;

said average memory utilization is less than a memory utilization threshold;

said average disk queue depth is equal to or greater than a disk queue depth threshold;

said LAN byte throughput is less than a LAN byte throughput threshold; and

,

said LAN packet throughput is less than a LAN packet throughput threshold.

16. The method as claimed in claim 12, wherein said step of applying a set of rules includes the computer implemented step of:

reporting a LAN bottleneck if;

said average CPU utilization is less than a CPU utilization threshold;

said average memory utilization is less than a memory utilization threshold;

said average disk queue depth is less than a disk queue depth threshold; and

,

said LAN byte throughput is equal to or greater than a LAN byte throughput threshold or said LAN packet throughput is equal to or greater than a LAN packet throughput threshold.

17. The method as claimed in claim 12, wherein said step of applying a set of rules includes the computer implemented step of:

reporting a system bottleneck if;

said average CPU utilization is equal to or greater than a CPU utilization threshold;

said average memory utilization is equal to or greater than a memory utilization threshold;

said average disk queue depth is equal to or greater than a disk queue depth threshold;

said LAN byte throughput is equal to or greater than a LAN byte throughput threshold; and

,

said LAN packet throughput is equal to or greater than a LAN packet throughput threshold.

18. The method of claim 11, wherein said server has a plurality of drives, said method further comprises:

checking which one of said plurality of drives has a lowest running average utilization; and

recommending that some of the work load be migrated to the drive with the lowest running average utilization.

19. A data processing system comprising:

a processor; and

a program code executed on said processor for predicting a hardware performance bottleneck, said program code including code for;

monitoring a hardware utilization parameters over selected time periods;

computing running averages of said hardware utilization parameter over said selected time periods;

determining a trend in said running averages;

predicting a bottleneck by comparing said trend to a threshold value for said hardware utilization parameter; and

outputting a prediction of said bottleneck to an output device.

20. The data processing system as claimed in claim 19, wherein said program code for determining said trend includes the step of applying linear regression to said running averages.

21. The data processing system as claimed in claim 19, wherein said program code for outputting said prediction includes code for outputting a recommendation to preempt said bottleneck.

22. The data processing system as claimed in claim 21, wherein:

said data processing system is a server within a LAN network; and

said utilization parameter is further selected from among LAN byte throughput and LAN packet throughput.

23. The data processing system as claimed in claim 19, further comprising:

a memory;

a disk having a queue dept; and

wherein said utilization parameter is selected from among average processor utilization, average memory utilization, and average disk queue depth.