Systems and methods for encoding knowledge for automated management of software application deployments

• US 7,490,073 B1
• Filed: 12/21/2005
• Issued: 02/10/2009
• Est. Priority Date: 12/21/2004
• Status: Active Grant

• Current Assignee: Citrix Systems, Inc. (Cloud Software Group)
• Original Assignee: Zenprise, Inc. (Cloud Software Group)
• Inventors: Hassan, Tanvir, Roach, Kelly Brian, Qureshi, Waheed, Page, Sekou
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): Kennedy; Adrian L
• Application Number:
  US11/316,214
• Time in Patent Office: 1,147 Days
• Field of Search: 706/50, 706/56, 706/57, 717/174
• US Class Current: 706/50
• CPC Class Codes: G06F 11/079 Root cause analysis, i.e. error or fault diagnosis in a hardware test environment G06F11/22; in a software test environment G06F11/36
• CPC Class Codes: G06N 5/048 Fuzzy inferencing

Abstract

A method of encoding knowledge is disclosed, which can be used to automatically detect problems in software application deployments. The method includes accessing a source of knowledge describing a problem known to occur in deployments of a particular software application, and which identifies a plurality of conditions associated with the problem. An encoded representation of the knowledge source is generated according to a predefined knowledge encoding methodology. The encoded representation is adapted to be applied automatically by a computer to analyze data representing a current state of a monitored deployment of the software application to detect whether the conditions and the problem exist therein. In various implementations, the encoded representation of the knowledge can include queries for deployment information, information concerning the relative importance of the conditions to a detection of the problem, and/or logical constructs for computing a confidence value in the existence of the problem and for determining whether to report the problem if some of the conditions are not true. The knowledge source can comprise a text document (such as a knowledge base article), a flowchart of a diagnostic troubleshooting method, and the like. Also disclosed are methods of at least partially automating the encoding process.

Claims

1. A method of encoding a diagnostic troubleshooting procedure for detecting problems in a deployment of a software application, comprising:

providing a human-readable diagnostic procedure configured to be used to detect problems in the deployment, the diagnostic procedure represented as a machine-readable tree comprising decision factors, at least some of the decision factors having “

yes” and

“

no”

paths;

encoding the decision factors into a machine-readable format, including encoding input and output information for each encoded decision factor;

using a computer-implemented tool to automatically convert the tree into a plurality of machine-readable rules, each of the rules corresponding to one of the problems, each of the rules comprising a specific navigation of a plurality of the decision factors and a plurality of “

yes”

or “

no”

paths of the tree, each of said navigations comprising a navigation from a root node of the tree to one of a plurality of terminal nodes of the tree; and

storing the rules in computer storage, in a format in which the rules can be used by a computer to automatically detect said problems in the deployment of the software application.

2. The method of claim 1, wherein the tree further comprises remedies, each remedy comprising a procedure for solving one of the problems, at least one of the rules comprising a specific navigation of decision factors of the tree ending in a remedy.

3. The method of claim 1, further comprising using the tool to automatically produce human-readable descriptions of each of the rules and its corresponding problem based on the tree and metadata in the decision factors.

4. The method of claim 1, further comprising using the tool to merge decision factors.

5. The method of claim 1, further comprising using the tool to produce rule metadata that can be used to compare one rule to another rule.

6. The method of claim 5, wherein the rule metadata can be used to evaluate a degree of relatedness of two of the rules.

7. A plurality of machine-readable rules generated according to the method of claim 1, represented in computer storage.

28. The method of claim 1, wherein the human-readable diagnostic procedure is provided in Microsoft Visio™

.

29. The method of claim 1, wherein encoding the decision factors comprises using a Microsoft Visio™

plugin.

30. The method of claim 1, the tool comprising a Microsoft Visio™

plugin.

8. A method of encoding knowledge that can be used to automatically detect problems in software application deployments, comprising:

accessing a source of knowledge describing a problem known to occur in deployments of a particular software application, and which identifies a plurality of conditions associated with the problem;

generating an encoded representation of the knowledge source according to a predefined knowledge encoding methodology, said encoded representation adapted to be applied automatically by a computer to analyze data representing a current state of a monitored deployment of said software application to detect whether said conditions and said problem exist therein; and

storing the encoded representation in computer storage, in a format in which the encoded representation can be used by the computer to automatically detect the problem in said monitored deployment of said software application;

wherein the encoded representation of the knowledge source includes logic for automatically generating, in connection with a detection of an occurrence of the problem, a confidence value representing a confidence that the detection is accurate;

wherein the encoded representation of the knowledge source specifies at least one query for affirmatively causing a collection of new telemetry data from said monitored deployment of the software application to obtain a new value of a time-varying state metric of the deployment when the encoded representation has not been previously used to evaluate said state metric during the detection of a particular one of the conditions, wherein the query is represented in the encoded representation, the encoded representation being configured to be used in combination with at least the value obtained via said query to evaluate whether the problem exists in the deployment.

9. The method of claim 8, wherein generating the encoded representation of the knowledge source comprises:

assigning different, non-zero numerical contribution weights corresponding to different ones of said conditions; and

storing the contribution weights in the encoded representation, such that the encoded representation gives more weight to at least one of the conditions than to at least one other condition for purposes of evaluating whether the problem exists.

10. The method of claim 8, wherein generating the encoded representation of the knowledge source comprises:

assigning different, non-zero numerical contribution weights corresponding to different ones of said conditions; and

storing the contribution weights in the encoded representation, such that the encoded representation gives more weight to at least one of the conditions than to at least one other condition for purposes of calculating a confidence value representative of the existence of the problem.

11. The method of claim 8, wherein the encoded representation of the knowledge source can be programmatically used to receive a condition confidence value representing a confidence that a particular condition exists, and use said condition confidence value to evaluate a confidence that the problem exists.

12. The method of claim 8, wherein generating the encoded representation of the knowledge source comprises:

designating at least one of the conditions as a required condition;

designating at least another one of the conditions as an optional condition; and

storing said designations as metadata attributes in the encoded representation, such that the encoded representation can be programmatically used to require the required condition for detection of the problem and to not require the optional condition for detection of the problem.

13. The method of claim 12, wherein the encoded representation includes a confidence calculation procedure for calculating a confidence in the existence of the problem, the confidence calculation procedure configured to increase the calculated confidence if the optional condition exists and to decrease the calculated confidence if the optional condition does not exist.

14. The method of claim 12, wherein the encoded representation includes a confidence calculation procedure for calculating a confidence in the existence of the problem, the confidence calculation procedure configured to increase the calculated confidence if the required condition exists and to compute a zero confidence if the required condition does not exist.

15. The method of claim 8, wherein the encoded representation of the knowledge source comprises a plurality of logic gates, wherein at least one of the gates is configured to produce a bead if one of the conditions exists within the deployment, and at least another of the gates is configured to receive at least one bead as an input and produce an output bead, the beads configured to be represented in computer storage, containing variable substitutions, and capable of propagating confidence information reflective of confidence levels that particular conditions exist within the deployment.

16. The method of claim 8, wherein the encoded representation of the knowledge source comprises a plurality of logic gates, wherein at least one of the gates is configured to produce a bead if a condition exists within the deployment, and at least another of the gates is configured to receive at least one bead as an input and produce an output bead, the beads configured to be represented in computer storage, containing variable substitutions, and capable of propagating time information representative of times that particular conditions are detected, such that the time information can be used to detect the problem.

17. The method of claim 8, wherein generating the encoded representation of the knowledge source comprises generating a plurality of predicates, at least some of which correspond to particular ones of said conditions.

18. The method of claim 8, wherein the encoded representation is configured to facilitate computerized analysis of at least one time-varying state metric representative of a dynamic state of the monitored deployment.

19. The method of claim 8, further comprising automatically modifying the encoded representation of the knowledge source using a stochastic back-propagation algorithm in response to feedback generated from application of the knowledge source at a deployment site of the software application.

20. An encoded representation of a knowledge source generated according to the method of claim 8 and represented in computer storage.

21. The method of claim 8, wherein the knowledge source comprises a text document.

22. The method of claim 21, wherein the text document comprises a knowledge base article created by a vendor of the software application.

23. The method of claim 8, wherein the knowledge source comprises a diagnostic method.

24. The method of claim 11, wherein generating the encoded representation of the knowledge source comprises:

specifying significance values for the conditions; and

storing the significance values in the encoded representation, such that the significance values can be programmatically used to decline to report the problem when (1) one or more of the conditions are found to be true, (2) the true conditions indicate that the problem is currently present in the deployment, and (3) a condition confidence value of at least one of the true conditions does not exceed the significance value of that particular condition.

25. The method of claim 12, wherein the encoded representation includes a confidence calculation procedure for calculating a confidence in the existence of the problem, the confidence calculation procedure configured to calculate a maximum confidence value if the required condition exists.

26. The method of claim 8, wherein the knowledge source comprises an oral recitation of knowledge.

31. The method of claim 8, wherein the encoded representation and the query are represented substantially in a logic programming language.

32. The method of claim 19, wherein generating the encoded representation of the knowledge source comprises:

assigning different, non-zero numerical contribution weights corresponding to different ones of said conditions; and

storing the contribution weights in the encoded representation, such that the encoded representation gives more weight to at least one of the conditions than to at least one other condition for purposes of evaluating whether the problem exists;

wherein modifying the encoded representation comprises adjusting one or more of said contribution weights.

33. The method of claim 19, wherein modifying the encoded representation comprises adjusting relative numerical weights given to one or more of the conditions for differently weighting importance levels of the conditions for purposes of (1) determining whether the problem exists in the deployment, or (2) determining the confidence value.

34. The method of claim 24, wherein the significance values are static numbers.

27. A method of encoding knowledge that can be used to automatically detect problems in software application deployments, comprising:

accessing a source of knowledge describing a problem known to occur in deployments of a particular software application, and which identifies a plurality of conditions associated with the problem;

generating an encoded representation of the knowledge source according to a predefined knowledge encoding methodology, said encoded representation adapted to be applied automatically by a computer to analyze state data of a monitored deployment of said software application to detect whether said conditions and said problem exist therein, wherein the encoded representation can be programmatically used to receive condition confidence values each representing a confidence that a particular one of the conditions is true;

providing within the encoded representation static significance values for the conditions, such that the significance values can be programmatically used to decline to report the problem when (1) one or more of the conditions are found to be true, (2) the true conditions indicate that the problem is currently present in the monitored deployment, and (3) a condition confidence value of at least one of the true conditions does not exceed the significance value of that particular condition; and

storing the encoded representation in computer storage, in a format in which the encoded representation can be used by the computer to automatically detect the problem in said monitored deployment of said software application.