Computer performance prediction using search technologies

• US 10,452,458 B2
• Filed: 01/13/2015
• Issued: 10/22/2019
• Est. Priority Date: 01/23/2014
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Seto, Tetsuo
• Primary Examiner(s): Kundu, Sujoy K
• Assistant Examiner(s): Anderson, L.
• Application Number:
  US14/596,126
• Publication Number:
  US 20150205690A1
• Time in Patent Office: 1,743 Days
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 11/323 Visualisation of programs or trace data
• CPC Class Codes: G06F 11/3409 for performance assessment
• CPC Class Codes: G06F 11/3442 for planning or managing the needed capacity
• CPC Class Codes: G06F 11/3447 Performance evaluation by modeling
• 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 2201/86 Event-based monitoring

Abstract

A computer monitoring system may predict near term and long term performance by comparing a segment of current time series data with previously observed time series to find matching segments. From a matching segment, a prediction of performance may be made by examining later observations in the time series. Each time series element may include a large number of parameters, and one mechanism for comparing segments may be treating the elements as multi-dimensional vectors and using cosine similarity for finding significant matches. A deployment mechanism may store time series segments in a searchable database, and search the database with a newly observed time series segment for matches.

Claims

1. A method for creating predictions using historical observation data, the method performed in a computing system comprising at least one computer processor, the method comprising:

receiving a first time series segment, the first time series segment comprising performance data observed while executing a first computer application;

using the first time series segment as a search basis, searching a historical database, the historical database comprising one or more time series, each of the one or more time series comprising historical performance data observed while executing the first computer application, each of the one or more time series comprising one or more time series segments;

identifying a second time series segment of a second time series from the historical database, the second time series comprising observations observed while executing the first computer application, the second time series segment identified as being similar to the first time series segment, similarity being determined by having a correlation value between the second time series segment and the first time series segment that meets or exceeds a predefined minimum correlation value;

identifying a third time series segment from within the second time series, the third time series segment being later in time in the second time series than the identified second time series segment; and

creating a prediction of future performance of the application based on the identified third time series segment.

2. The method of claim 1 further comprising:

identifying a fourth time series segment from the historical database, the fourth time series segment being similar to the first time series segment, similarity being determined by having a correlation value between the fourth time series segment and the first time series segment that meets or exceeds a second predefined minimum correlation value;

identifying a fifth time series segment from the historical database, the fifth time series being later in sequence from the fourth time series segment; and

creating the prediction based on the third time series segment and the fifth time segment.

3. The method of claim 2, said prediction being created by at least weighting said third time series with a first weight and weighting said fifth time series with a second weight.

4. The method of claim 2, said historical database comprising a plurality of time series, at least one of said time series being observations while executing said first computer application on a first execution platform and at least a second one of said time series being observations while executing said first computer application on a second execution platform.

5. The method of claim 4, said third time series segment being observations from said first execution platform and said fifth time series segment being observations from said second execution platform.

6. The method of claim 4, said historical database further comprising at least one time series being observations while executing a second computer application.

7. The method of claim 6, said third time series segment being observations from said first computer application and said fifth time series segment being observations from said second computer application.

8. The method of claim 7, said first computer application being a version of said second computer application.

9. The method of claim 1, said performance data comprising performance data for a first function and a second function within said first application.

10. The method of claim 9, said performance data comprising aggregated performance data for a time interval in said time series.

11. The method of claim 1, wherein the method further includes the computing system performing at least one of triggering an alert, performing a corrective action or performing a load management process based on the prediction of future performance.

12. A system comprising:

a processor;

a database comprising a plurality of time series representing performance observations from tracing a first application;

a search engine that performs at least;

receiving a first time series segment, the first time series segment comprising performance data observed while executing the first computer application;

using the first time series segment as a search basis, searching a historical database, the historical database comprising one or more time series, each of the one or more time series comprising historical performance data observed while executing the first computer application, each of the one or more time series comprising one or more time series segments;

identifying a second time series segment of a second time series from the historical database, the second time series comprising observations observed while executing the first computer application, the second time series segment identified as being similar to the first time series segment, similarity being determined by having a correlation value between the second time series segment and the first time series segment that meets or exceeds a predefined minimum correlation value; and

returning the second time series segment;

an analysis engine that performs at least;

identifying a third time series segment from within the second time series, the third time series segment being later in time in the second time series than the identified second time series segment; and

creating a prediction of future performance of the application based on the identified third time series segment.

13. The system of claim 12, the search engine that further:

identifies a fourth time series segment from the historical database, the fourth time series segment being similar to the first time series segment, similarity being determined by having a correlation value between the fourth time series segment and the first time series segment that meets or exceeds a second predefined minimum correlation value;

the analysis engine that further;

identifies a fifth time series segment from the historical database, the fifth time series being later in sequence from the fourth time series segment; and

creates the prediction based on the third time series segment and the fifth time segment.

14. The system of claim 13, said prediction being created by at least weighting said third time series with a first weight and weighting said fifth time series with a second weight.

15. The system of claim 13, said historical database comprising a plurality of time series, at least one of said time series being observations while executing said first computer application on a first execution platform and at least a second one of said time series being observations while executing said first computer application on a second execution platform.

16. The system of claim 15, said third time series segment being observations from said first execution platform and said fifth time series segment being observations from said second execution platform.

17. The system of claim 15, said historical database further comprising at least one time series being observations while executing a second computer application.

18. The system of claim 17, said third time series segment being observations from said first computer application and said fifth time series segment being observations from said second computer application.

19. The system of claim 18, said first computer application being a version of said second computer application.

20. The system of claim 12, said performance data comprising performance data for a first function and a second function within said first application.

21. The system of claim 20, said performance data comprising aggregated performance data for a time interval in said time series.