Selecting instrumentation points for an application

• US 9,009,680 B2
• Filed: 12/18/2006
• Issued: 04/14/2015
• Est. Priority Date: 11/30/2006
• Status: Active Grant

• Current Assignee: CA, Inc. (d/b/a CA Technologies) (Broadcom, Inc.)
• Original Assignee: CA, Inc. (d/b/a CA Technologies) (Broadcom, Inc.)
• Inventors: Cobb, Jeffrey R., Puryear, Daryl L., Thio, Ling
• Primary Examiner(s): Zhen, Li B
• Assistant Examiner(s): BODDEN, EVRAL E
• Application Number:
  US11/612,058
• Publication Number:
  US 20080148039A1
• Time in Patent Office: 3,039 Days
• Field of Search: None
• US Class Current: 717/130
• CPC Class Codes: G06F 11/3612 by runtime analysis performance monitoring G06F11/3466
• CPC Class Codes: G06F 11/3624 by performing operations on the source code, e.g. via a compiler

Abstract

Instrumentation points are selected for an application by running the application with comprehensive instrumentation of its components in a development mode. The application can be run by a human operator and/or load simulator which provides requests to the application. The instrumented components are monitored as the application runs. A subset of the components is selected based on criteria such as an order in which the instrumented components are invoked, whether resource utilization, such as consumption of processor cycles, exceeds a threshold, or a frequency with which components are called or call other components, and only that subset is instrumented in a production mode of the application. In one approach, the subset includes components which are invoked when traffic to/from the application matches a pattern provided by an interaction model. As a result, relevant instrumentation points can be identified.

Claims

1. A computer-implemented method for selecting instrumentation points for an application, comprising:

performing a static analysis of components of an application, the performing the static analysis comprises determining which of the components are worthy of instrumenting, the determining which of the components are worthy of instrumenting comprises determining which of the components are of at least one specific type which is of interest;

instrumenting the components which are determined to be worthy of instrumenting based on the static analysis to provide a first set of components, each component of the first set of components is instrumented with a probe which is initially made active so that the probe will be tripped and send component data to an agent when the component of the first set of components is invoked;

in one iteration of a filtering process;

running the application with the first set of components while monitoring the first set of components based on probes of the first set of components,

based on the monitoring of the first set of components, for each component of the first set of components;

assigning a first score based on a first selection criterion, assigning a second score based on a second selection criterion, and providing a total score based on the first score and the second score and

based on the total scores, distinguishing a first subset of components of the first set of components for which active instrumentation is to be maintained from a second subset of components of the first set of components for which active instrumentation is not to be maintained; and

in another iteration of the filtering process;

running the application with the first subset of components while keeping probes for the first subset of components active and making probes for the second subset of components inactive, where the probes for the second subset of components remain in the application, and do not send component data to an agent, in the another iteration, and while monitoring the first subset of components based on the probes of the components of the first set of components,

based on the monitoring the first subset of components, assessing the first subset of components using another selection criterion, and

based on the assessing the first subset of components, distinguishing a subset of the first subset of components for which active instrumentation is to be maintained in subsequent running of the application from a subset of the first subset of components for which active instrumentation is not to be maintained in the subsequent running of the application.

2. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises parsing HyperText Transport Protocol (HTTP) requests to the application to obtain parameters of the HTTP requests, comparing the parameters of the HTTP requests to at least one pattern associated with an interaction model for the application to determine at least one of the HTTP requests which corresponds to the at least one pattern associated with the interaction model for the application, and selecting the first subset of components by selecting components of the first set of components which are invoked by the at least one of the HTTP requests.

3. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components based on an order in which the components of the first set of components are invoked during the running of the application with the first set of components.

4. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components for which a resource utilization exceeds a threshold during the running of the application with the first set of components.

5. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components for which at least one of a consumption of processor cycles or an execution time exceeds a threshold during the running of the application with the first set of components.

6. The computer-implemented method of claim 1, further comprising:

determining an instrumentation overhead of each component of the first set of components during the running of the application with the first set of components, based on a ratio of central processing unit (CPU) cycles consumed by instrumentation of each component of the first set of components to CPU cycles consumed when each component of the first set of components is invoked, the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components for which the instrumentation overhead is determined to be below a threshold.

7. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components which are called with a frequency above an associated threshold during the running of the application with the first set of components.

8. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components which are called by a number of other components which exceeds an associated threshold during the running of the application with the first set of components.

9. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components for which a number of multiple instances which exceeds a threshold are in use simultaneously during the running of the application with the first set of components.

10. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components which at least one of;

provide an entry point into another component, make method calls to other components or have a length which exceeds a length threshold.

11. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components which have a variance in execution time which exceeds a threshold during the running of the application with the first set of components.

12. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by determining which of the components of the first set of components is first invoked when processing a request during the running of the application with the first set of components, the first subset of components including the component of the first set of components which is first invoked.

13. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by determining which of the components of the first set of components is last invoked when processing a request during the running of the application with the first set of components, the first subset of components including the component of the first set of components which is last invoked.

14. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by determining which of the components of the first set of components call an application programming interface during the running of the application with the first set of components, the first subset of components including the components of the first set of components which call an application programming interface.

17. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components which call other components with a frequency above an associated threshold during the running.

18. The computer-implemented method of claim 1, wherein:

the distinguishing during the one iteration comprises selecting the first subset of components by selecting components of the first set of components which call a number of other components which exceeds an associated threshold during the running.

19. The computer-implemented method of claim 1, wherein:

the at least one specific type which is of interest comprises at least one of a servlet, a Java Server Page, an Enterprise Java Bean, a Java Database Connectivity component or a .NET component.

20. The computer-implemented method of claim 1, wherein:

the determining which of the components are worthy of instrumenting comprises parsing and analyzing code of the application to detect which of the components is associated with a specific name of a class of interest.

21. The computer-implemented method of claim 1, wherein:

the determining which of the components are worthy of instrumenting comprises parsing and analyzing code of the application to detect which of the components is associated with a specific name of a direct or indirect super class of interest.

22. The computer-implemented method of claim 1, wherein:

the determining which of the components are worthy of instrumenting comprises parsing and analyzing code of the application to detect which of the components is associated with a specific name of a directly- or indirectly-implemented interface of interest.

23. The computer-implemented method of claim 1, wherein:

the determining which of the components are worthy of instrumenting comprises parsing and analyzing code of the application to detect which of the components is associated with a specific name of a method of interest.

24. The computer-implemented method of claim 1, wherein:

the another selection criterion is more exclusive than the first selection criterion.

25. The computer-implemented method of claim 24, wherein:

the first selection criterion comprises one execution time; and

the another selection criterion comprises another execution time which is greater than the one execution time.

26. The computer-implemented method of claim 24, wherein:

the first selection criterion comprises one variance in execution time; and

the another selection criterion comprises another variance in execution time.

27. The computer-implemented method of claim 24, wherein:

the first selection criterion comprises one number of components called; and

the another selection criterion comprises another number of components called.

15. At least one processor readable memory having processor readable code embodied thereon, at least one processor is programmed by the processor readable code and performs a method comprising:

performing a static analysis of components of an application, the performing the static analysis comprises determining which of the components are worthy of instrumenting, the determining which of the components are worthy of instrumenting comprises determining which of the components are of at least one specific type which is of interest;

instrumenting the components which are determined to be worthy of instrumenting based on the static analysis to provide a first set of components, each component of the first set of components is instrumented with a probe which is initially made active so that the probe will be tripped and send component data to an agent when the component of the first set of components is invoked;

running the application with the first set of components with the probes which are made active;

monitoring the first set of components during the running based on the probes which are made active; and

based on the monitoring;

for each component of the first set of components;

assigning a first score based on a first selection criterion, assigning a second score based on a second selection criterion, and providing a total score based on the first score and the second score,

based on the total scores, distinguishing a first subset of components of the first set of components for which active instrumentation is to be maintained in subsequent running of the application, from a second subset of components of the first set of components for which active instrumentation is not to be maintained in the subsequent running of the application,

keeping probes for the first subset of components active in the subsequent running of the application, and

making probes for the second subset of components inactive in the subsequent running of the application, where the probes for the second subset of components remain in the application and do not send component data to an agent, in the subsequent running of the application.

28. The at least one processor readable memory of claim 15, wherein:

the first score has a different weight than the second score based on a relative importance of the first selection criterion to the second selection criterion.

16. A system for selecting instrumentation points for an application, comprising:

a storage device; and

a processor in communication with the storage device, the processor;

carries out a performance of a static analysis of components of an application, the performance of the static analysis comprises a determination of a plurality of the components which are worthy of instrumenting;

divides the plurality of the components which are worthy of instrumenting into different sets of components comprising a first set of components and a second set of components;

in one iteration of an iterative process;

instruments each component of the first set of components with a probe which is made active so that the probe will be tripped and send component data to an agent when the component is invoked,

runs the application with the first set of components with the probes which are made active, while the second set of components does not have probes which are made active,

monitors the first set of components during the running based on the probes of the first set of components which are made active, and

based on the monitoring of the first set of components, for each component of the first set of components;

assigns a first score based on a first selection criterion, assigns a second score based on a second selection criterion, and provides a total score based on the first score and the second score, and based on the total scores, distinguishes a first subset of components of the first set of components which meets a selection criterion from a second subset of components of the first set of components which does not meet the selection criterion;

in another iteration of the iterative process;

instruments each component of the second set of components with a probe which is made active so that the probe will be tripped and send component data to an agent when the component is invoked,

runs the application with the second set of components with the probes which are made active, while the first set of components does not have probes which are made active,

monitors the second set of components during the running based on the probes of the second set of components which are made active, and

based on the monitoring of the second set of components, distinguishes a first subset of components of the second set of components which meets the selection criterion from a second subset of components of the second set of components which does not meet the selection criterion; and

based on the one iteration and the another iteration, configures the application to subsequently run with probes which are active for the first subset of components of the first set of components and for the first subset of components of the second set of components, and to run without probes which are active for the second subset of components of the first set of components and for the second subset of components of the second set of components.

29. The system of claim 16, wherein:

in the one iteration, the second set of components is not instrumented with probes; and

in the another iteration, the probes of the first set of components, including probes of the first subset of components of the first set of components, are removed; and

to configure the application to subsequently run with probes which are active for the first subset of components of the first set of components, the processor adds the probes of the first subset of components of the first set of components back into the first subset of components of the first set of components.

30. The system of claim 16, wherein:

the processor divides the plurality of the components which are worthy of instrumenting into the different sets of components comprising the first set of components and the second set of components according to a first range of alphabetical names of the first set of components and a second range of alphabetical names of the second set of components.

31. A computer-implemented method for selecting instrumentation points for an application, comprising:

performing a static analysis of components of an application, the performing the static analysis comprises determining which of the components are worthy of instrumenting, the determining which of the components are worthy of instrumenting comprises determining which of the components are of at least one specific type which is of interest;

instrumenting the components which are determined to be worthy of instrumenting based on the static analysis to provide a first set of components, each component of the first set of components is instrumented with a probe which is initially made active so that the probe will be tripped and send component data to an agent when the component of the first set of components is invoked;

in one iteration of a filtering process;

running the application with the first set of components while monitoring the first set of components based on probes of the first set of components,

based on the monitoring of the first set of components, assessing the first set of components using one selection criterion, and

based on the assessing the first set of components using the one selection criterion, distinguishing a first subset of components of the first set of components for which active instrumentation is to be maintained from a second subset of components of the first set of components for which active instrumentation is not to be maintained; and

in another iteration of the filtering process;

running the application with the first subset of components while keeping probes for the first subset of components active and making probes for the second subset of components inactive, where the probes for the second subset of components remain in the application, and do not send component data to an agent, in the another iteration, and while monitoring the first subset of components based on the probes of the components of the first set of components,

based on the monitoring the first subset of components, for each component of the first subset of components;

assigning a first score based on a first selection criterion, assigning a second score based on a second selection criterion, and providing a total score based on the first score and the second score; and

based on the total scores, distinguishing a subset of the first subset of components for which active instrumentation is to be maintained in subsequent running of the application from a subset of the first subset of components for which active instrumentation is not to be maintained in the subsequent running of the application.