Continuous system health indicator for managing computer system alerts

• US 8,458,530 B2
• Filed: 09/21/2010
• Issued: 06/04/2013
• Est. Priority Date: 09/21/2010
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Kini, Prashanth, Tauriello, Richard, Campos, Marcos M., Milenova, Boriana, Lubomirsky, Andres V., Sum, Charlie
• Primary Examiner(s): Manoskey, Joseph D
• Application Number:
  US12/886,891
• Publication Number:
  US 20120072780A1
• Time in Patent Office: 987 Days
• Field of Search: 714/25, 714/26, 714/33, 714/37, 714/46, 714/47.1, 714/47.2, 714/47.3, 714/57
• US Class Current: 714/47.1
• CPC Class Codes: G06F 11/328 Computer systems status display G06F11/327 takes precedence
• CPC Class Codes: G06F 11/3419 by assessing time
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis
• CPC Class Codes: G06F 11/3466 Performance evaluation by tracing or monitoring
• CPC Class Codes: G06F 2201/865 Monitoring of software

Abstract

A method is provided for detecting when users are being adversely impacted by poor system performance. A system health indicator is determined that is based on the amount of work that is blocked waiting for each of a set of an external events and combined with a heuristic that is based on the number of users waiting for the work to complete. The system health indicator is compared to a threshold such that an alert is generated when the system health indicator crosses the threshold. However, the system health indicator is designed so that an alert is only generated when a significant user base is or will in the near future experience a problem with the system. Furthermore, the system health indicator is designed to vary smoothly to maintain its suitability for the application of predictive technology.

Claims

1. A computer-implemented method for measuring health of a computer system over a time interval comprising steps of:

determining, based on an activity measurement value and a set of time measurement values, a health indicator value that indicates the health of the computer system over the time interval;

wherein the activity measurement value is based on a number of sessions for which the computer system is actively performing work during the time interval;

wherein each time measurement value of the set of time measurement values is associated with a wait class of a plurality of wait classes and represents an amount of time during which said work performed by the computer system spends waiting in one of a plurality of wait classes;

determining an activity weighting factor based on a first scaling function of the activity measurement value, wherein the first scaling function scales the activity measurement value to a value between 0 and a first pre-determined upper bound;

determining an amount of work waiting on each wait class of the plurality of wait classes;

determining, based on a second scaling function of the amount of work waiting on said each wait class of the plurality of wait classes, a set of user impact weighting factors associated with said each wait class, wherein the second scaling function scales the amount of work waiting on said each wait class to a number between 0 and a second pre-determined upper bound; and

one or more computing devices computing the health indicator value over the time interval as a function of the set of time measurement values, the amount of work waiting on said each wait class of the plurality of wait classes, the set of user impact weighting factors, and the activity weighting factor.

2. The method of claim 1, wherein the activity measurement value is a number of active user sessions.

3. The method of claim 1, wherein each time measurement value of the set of time measurement values is a percentage of time spent in a corresponding wait class of the plurality of wait classes.

4. The method of claim 1, wherein the plurality of wait classes include user I/O, network I/O, and waiting for locks.

5. The method of claim 1, wherein said amount of work corresponding to said each wait class is a number of user sessions waiting in the corresponding said each wait class.

6. The method of claim 1, the steps further comprising sending an alert when the health indicator value exceeds a threshold.

7. The method of claim 6, wherein an initial threshold value is determined during a machine learning period for said computer system.

8. The method of claim 1, wherein the threshold is dynamically adjusted based on historical health indicator values determined during the operation of said computer system.

9. The method of claim 1, wherein the activity measurement value and set of time measurement values are collected at a collection time interval of a first duration and aggregated over the time interval, wherein the time interval is an aggregation time interval of a second duration;

wherein the second duration is longer than the first duration, and the activity measurement value and set of time measurement values collected during the collection time interval are aggregated over the aggregation time interval.

10. The method of claim 9, wherein aggregating the activity measurement value and the set of time measurement values over the aggregation time interval is performed as a rolling average over a data collected at a particular number of most-recent time intervals.

11. The method of claim 1, wherein the system health indicator value is computed as a product of the activity weighting factor and a total adjusted wait time, wherein the total adjusted wait time is a sum of adjusted wait times for each wait class;

and

wherein each adjusted wait time for said each wait class is a product of a time measurement value associated with said each wait class and a user impact value associated with said each wait class.

12. A non-transitory computer-readable medium storing one or more instructions which when executed by one or more processors cause the one or more processors to perform steps comprising:

determining, based on an activity measurement value and a set of time measurement values, a health indicator value that indicates the health of the computer system over the time interval;

wherein the activity measurement value is based on a number of sessions for which the computer system is actively performing work during the time interval;

wherein each time measurement value of the set of time measurement values is associated with a wait class of a plurality of wait classes and represents an amount of time during which said work performed by the computer system spends waiting in one of a plurality of wait classes;

determining an activity weighting factor based on a first scaling function of the activity measurement value, wherein the first scaling function scales the activity measurement value to a value between 0 and a first pre-determined upper bound;

determining an amount of work waiting on each wait class of the plurality of wait classes;

determining, based on a second scaling function of the amount of work waiting on said each wait class of the plurality of wait classes, a set of user impact weighting factors associated with said each wait class, wherein the second scaling function scales the amount of work waiting on said each wait class to a number between 0 and a second pre-determined upper bound; and

one or more computing devices computing the health indicator value over the time interval as a function of the set of time measurement values, the amount of work waiting on said each wait class of the plurality of wait classes, the set of user impact weighting factors, and the activity weighting factor.

13. The non-transitory computer-readable medium of claim 12, wherein the activity measurement value is a number of active user sessions.

14. The non-transitory computer-readable medium of claim 12, wherein each time measurement value of the set of time measurement values is a percentage of time spent in a corresponding wait class of the plurality of wait classes.

15. The non-transitory computer-readable medium of claim 12, wherein the plurality of wait classes include user I/O, network I/O, and waiting for locks.

16. The non-transitory computer-readable medium of claim 12, wherein said amount of work corresponding to said each wait class is a number of user sessions waiting in the corresponding said each wait class.

17. The non-transitory computer-readable medium of claim 12, the steps further comprising sending an alert when the health indicator value exceeds a threshold.

18. The non-transitory computer-readable medium of claim 17, wherein an initial threshold value is determined during a machine learning period for said computer system.

19. The non-transitory computer-readable medium of claim 12, wherein the threshold is dynamically adjusted based on historical health indicator values determined during the operation of said computer system.

20. The non-transitory computer-readable medium of claim 12, wherein the activity measurement value and set of time measurement values are collected at a collection time interval of a first duration and aggregated over the time interval, wherein the time interval is an aggregation time interval of a second duration;

wherein the second duration is longer than the first duration, and the activity measurement value and set of time measurement values collected during the collection time interval are aggregated over the aggregation time interval.

21. The non-transitory computer-readable medium of claim 20, wherein aggregating the activity measurement value and the set of time measurement values over the aggregation time interval is performed as a rolling average over a data collected at a particular number of most-recent time intervals.

22. The non-transitory computer-readable medium of claim 12, wherein the system health indicator value is computed as a product of the activity weighting factor and a total adjusted wait time, wherein the total adjusted wait time is a sum of adjusted wait times for each wait class;

and

wherein each adjusted wait time for said each wait class is a product of a time measurement value associated with said each wait class and a user impact value associated with said each wait class.