Recommending a Suspicious Component in Problem Diagnosis for a Cloud Application

• US 20150199226A1
• Filed: 01/07/2015
• Published: 07/16/2015
• Est. Priority Date: 01/16/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ye, Meng, Yu, Tao, Wu, Hai S.
• Granted Patent: US 9,477,544 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/0709 in a distributed system consisting of a plurality of standalone computer nodes, e.g. clusters, client-server systems
• 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: H04L 41/0677 Localisation of faults
• CPC Class Codes: H04L 41/5019 Ensuring fulfilment of SLA
• CPC Class Codes: H04L 43/0829 Packet loss

Abstract

The present invention discloses a method and an apparatus for recommending a suspicious component in problem diagnosis for a cloud application. In the method, firstly a graph data model representing a hierarchical structure of the cloud application is constructed, wherein the graph data model comprises an application node representing the cloud application, a plurality of component nodes representing a plurality of components of the cloud application, and inter-node lines indicating inter-node relationships. Then real-time information of the cloud application is obtained, in response to detecting performance degradation of the cloud application. Impact degrees of the plurality of component nodes on the performance degradation of the cloud application is obtained based on the constructed graph data model and the obtained real-time information, and a suspicious component sequence is generated according to the impact degrees of the plurality of component nodes.

Claims

1. A method for recommending a suspicious component in problem diagnosis for a cloud application, comprising:

constructing a graph data model representing a hierarchical structure of the cloud application, wherein the graph data model comprises an application node representing the cloud application, a plurality of component nodes representing a plurality of components of the cloud application, and inter-node lines indicating inter-node relationships;

obtaining real-time information of the cloud application, in response to detecting performance degradation of the cloud application;

obtaining impact degrees of the plurality of component nodes on the performance degradation of the cloud application, based on the graph data model and the real-time information; and

generating a suspicious component sequence, according to the impact degrees of the plurality of component nodes.

2. The method according to claim 1, wherein the inter-node relationships comprise at least one of the followings:

a parent-child relationship for inter-layer nodes, a prior/next relationship or a peer relationship for intra-layer nodes.

3. The method according to claim 1, wherein the obtaining of the impact degrees of the plurality of component nodes on the performance degradation of the cloud application based on the graph data model and the real-time information comprises:

analyzing the real-time information to obtain a problem transaction with a worst performance index in the cloud application;

determining a path comprising component nodes involved in the problem transaction as a critical path, according to the graph data model, wherein one endpoint of the critical path is the application node; and

obtaining the impact degrees of the plurality of component nodes on the critical path, as the impact degrees of the plurality of component nodes on the performance degradation of the cloud application.

4. The method according to claim 3, wherein the analyzing of the real-time information to obtain the problem transaction with the worst performance index in the cloud application comprises:

extracting, from the real-time information, transaction information of a proxy server node being one of the plurality of component nodes; and

determining the problem transaction according to the transaction information of the proxy server node.

5. The method according to claim 4, wherein the worst performance index is response time.

6. The method according to claim 5, wherein the determining of the problem transaction according to the transaction information of the proxy server node comprises:

obtaining the response time of each transaction from the transaction information;

comparing the response time of the respective transactions with each other; and

determining the transaction with the longest response time as the problem transaction.

7. The method according to claim 5, wherein the determining of the problem transaction according to the transaction information of the proxy server node comprises:

obtaining the response time of each transaction from the transaction information;

calculating a ratio of a difference between the response time of the respective transactions and its reference response time to the reference response time; and

determining the transaction with the largest ratio as the problem transaction.

8. The method according to claim 3, wherein the obtaining of the impact degrees of the plurality of component nodes on the critical path comprises:

calculating, for each of the plurality of component nodes, a number of paths passing through the critical path of at least one path from the component node to the application node, as a first value of the component node; and

calculating a distance value of the at least one path from the component node to the application node, a maximum distance value being a second value of the component node;

wherein the impact degree comprises the first value acid the second value.

9. The method according to claim 3, wherein the obtaining of impact degrees of the plurality of component nodes on the critical path comprises:

selecting at least one related component node having the inter-node relationship with the component nodes of the critical path;

calculating, for each component node in a node set comprising the component nodes of the critical path and the at least one related component node, a number of paths passing through the critical path of at least one path from the component node to the application node, as a first value of the component node; and

calculating a distance value of the at least one path from the component node to the application node, a maximum distance value being a second value of the component node;

wherein the impact degree comprises the first value and the second value.

10. The method according to claim 1, wherein the generating of the suspicious component sequence, according to the impact degrees of the plurality of component nodes comprises:

sorting the plurality of component nodes in descending order, according to the first values, to generate a first component sequence; and

sorting the component nodes with the same first value of the first component sequence in descending order, according to the second values, to generate a second component sequence, the second component sequence being the suspicious component sequence.

11. The method according to claim 1, wherein the generating of a suspicious component sequence, according to the impact degrees of the plurality of component nodes comprises:

extracting, from the real-time information, running status information of the plurality of component nodes;

determining priority component nodes with the running status information exceeding a predefined threshold;

sorting the priority component nodes in descending order, according to the first values, to generate a first component sequence;

sorting the priority component nodes with the same first value of the first component sequence, according to the second values, to generate a second component sequence;

sorting other component nodes except the priority component nodes in descending order, according to the first values, to generate a third component sequence;

sorting the component nodes with the same first value of the third component sequence in descending order, according to the second values, to generate a fourth component sequence; and

combining the second component sequence and the fourth component sequence in an order that the second component sequence precedes the fourth component sequence, to generate the suspicious component sequence.

12. An apparatus for recommending a suspicious component in problem diagnosis for a cloud application, comprising:

processor; and

a memory coupled to the processor, wherein the memory comprises instructions which, when executed by the processor cause the processor to;

construct a graph data model representing a hierarchical structure of the cloud application, wherein the graph data model comprises an application node representing the cloud application, a plurality of component nodes representing a plurality of components of the cloud application, and inter-node lines indicating inter-node relationships;

obtain real-time information of the cloud application, in response to detecting performance degradation of the cloud application;

obtain impact degrees of the plurality of component nodes on the performance degradation of the cloud application, based on the graph data model and the real-time information; and

generate a suspicious component sequence, according to the impact degrees of the plurality of component nodes.

13. The apparatus according to claim 12, wherein the instructions to obtain the impact degrees of the plurality of component nodes on the performance degredation of the cloud application based on the graph data model and the real-time information further cause the processor to:

analyze the real-time information to obtain a problem transaction with a worst performance index in the cloud application;

determine a path comprising component nodes involved in the problem transaction as a critical path, according to the graph data model, wherein one endpoint of the critical path is the application node; and

obtain the impact degrees of the plurality of component nodes on the critical path, as the impact degrees of the plurality of component nodes on the performance degradation of the cloud application.

14. The apparatus according to claim 13, wherein the instructions to analyze of the real-time information to obtain the problem transaction with the worst performance index in the cloud application further cause the processor to:

extract, from the real-time information, transaction information of a proxy server node being one of the plurality of component nodes; and

determine the problem transaction according to the transaction information of the proxy server node.

15. The apparatus according to claim 14, wherein the worst performance indicator is response time.

16. The apparatus according to claim 15, wherein the instructions to determine of the problem transaction according to the transaction information of the proxy server node further cause the processor to:

obtain the response time of each transaction from the transaction information;

compare the response time of the respective transaction with each other, and

determine the transaction with the longest response time as the problem transaction.

17. The apparatus according to claim 15, wherein the instructions to determine of the problem transaction according to the transaction information of the proxy server node further cause the processor to:

obtain the response time of each transaction from the transaction information;

calculate a ratio of a difference between the response time of the respective transactions and its reference response time to the reference response time and

determine the transaction with the largest ratio as the problem transaction.

18. The apparatus according to claim 13, wherein the plurality of component nodes on the critical path further cause the processor to:

calculate, for each of the plurality of component nodes, a number of paths passing through the critical path of at least one path from the component node to the application node as a first value of the component node; and

calculate a distance value of the at least one path from the component node to the application node, wherein a maximum distance value is taken as a second value of the component node;

wherein the impact degree comprises the first value and the second value.

19. The apparatus according to claim 13, wherein the instructions to obtain of the impact degrees of the plurality of component nodes on the critical path further cause the processor to:

select at least one related component node having the inter-node relationship with the component nodes of the critical path;

calculate, for each component node in a node set comprising the component nodes of the critical path and the at least one related component node, a number of paths passing through the critical path of at least one path from the component node to the application node, as a first value of the component node; and

calculate a distance value of the at least one path from the component node to the application node, wherein a maximum distance value is taken as a second value of the component node;

wherein the impact degree comprises the first value and the second value.

20. The apparatus according to claim 12, wherein the instructions to generate the suspicious component sequence, according to the impact degrees of the plurality of component nodes further causes the processor to:

sort the plurality of component nodes in descending order, according to the first values, to generate a first component sequence; and

sort the component nodes with the same first value of the first component sequence in descending order, according to the second values, to generate a second component sequence, wherein the second component sequence is taken as the suspicious component sequence.

21. The apparatus according to claim 12, wherein the instructions to generate the suspicious component sequence, according to the impact degrees of the plurality of component nodes further causes the processor to:

extract running stat s information of the plurality of component nodes from the real-time information;

determine priority component nodes with the running status information exceeding a predefined threshold;

sort the priority component nodes in descending order according to the first values to generate a first component sequence, and sort other component nodes except the priority component nodes in descending order according to the first values to generate a third component sequence;

sort the priority component nodes with the same first value of the first component sequence according to the second values to generate a second component sequence, and sort the component nodes with the same first value of the third component sequence in descending order according to the second values to generate a fourth component sequence; and

combine the second component sequence and the fourth component sequence in an order that the second component sequence precedes the fourth component sequence, to generate the suspicious component sequence.

22. A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to:

construct a graph data model representing a hierarchical structure of the cloud application, wherein the graph data model comprises an application node representing the cloud application, a plurality of component nodes representing a plurality of components of the cloud application, and inter-node lines indicating inter-node relationships;

obtain real-time information of the cloud application, in response to detecting performance degradation of the cloud application;

obtain impact degrees of the plurality of component nodes on the performance degradation of the cloud application, based on the graph data model and the real-time information; and

generate a suspicious component sequence, according to the impact degrees of plurality of component nodes.