Apparatus and method for managing network system

• US 8,769,339 B2
• Filed: 01/05/2011
• Issued: 07/01/2014
• Est. Priority Date: 01/12/2010
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Sonoda, Masataka, Wada, Yuji, Sekiguchi, Atsuji
• Primary Examiner(s): GUYTON, PHILIP A
• Application Number:
  US12/985,019
• Publication Number:
  US 20110173500A1
• Time in Patent Office: 1,273 Days
• Field of Search: 714/26
• US Class Current: 714/26
• CPC Class Codes: H04L 41/064 involving time analysis

Abstract

An apparatus for managing a network system including a plurality of components, the apparatus includes a memory that stores component type data of each component of the plurality of components, component relation data including relation information indicating a pair of components related to each other in the network system and error history data including error information of respective error components in the plurality of components. The apparatus includes a processor that executes a procedure including extracting a pair of component type data as a relation class candidate on the basis of the component type data of a pair of error components indicated by the error information in the error history data, the pair of error components being indicated by the relation information.

Claims

1. An apparatus for managing a network system including a plurality of components, the apparatus comprising:

a memory that stores;

component type data of each component of the plurality of components,

component relation data including relation information indicating a pair of components related to each other in the network system, the relation information including a first component type data of a first component requesting a process of the component and a second component type data of a second component accepting a request of the process requested by the first component, and

error history data indicating error information of the plurality of components; and

a processor coupled to the memory and which is configured to;

extract candidates of a relation class indicating a propagating direction of an error between components, based on the component type data, component relation data, and the error history data,

determine the relation class among the candidates based on the first component type data and the second component type data,

determine an investigation range by predetermining a specified hop value and decrementing the hop value every time the relation class is tracked, and

estimate a point where a failure causing the error has occurred by tracking the relation class within the investigation range.

2. The apparatus according to claim 1, wherein the processor is configured to count, for each combination of the pair of component type data, an appearance number of the candidates including the identical combination of the pair of component type data to the each combination;

and

wherein the determine the relation class includes determining the relation class from among the candidates on the basis of the counted appearance numbers.

3. The apparatus according to claim 1, wherein the network system is a virtual network system and the each of the plurality of components is a physical machine, a virtual machine or a service.

4. A method of managing a network system including a plurality of components, the method comprising:

obtaining component type data of each component of the plurality of components, component relation data including relation information indicating a pair of components related to each other in the network system, the relation information including a first component type data of a first component requesting a process of the component and a second component type data of a second component accepting a request of the process requested by the first component, and error history data indicating error information of the plurality of components,

extracting, by the processor, candidates of a relation class indicating a propagating direction of an error between components, based on the component toe data, component relation data and the error history data,

determining, by the processor, the relation class among the candidates based on the first component type data and the second component type data,

determining, by the processor, an investigation range by predetermining a specified hop value and decrementing the hop value every time the relation class is tracked, and

estimating, by a processor, a point where a failure causing the error has occurred by tracking the relation class within the investigation range.

5. The method according to claim 4, further comprising:

counting, for each combination of the pair of component type data, an appearance number of the relation class candidates including the identical combination of the pair of component type data to the each combination;

wherein

the determining of the relation class includes determining the relation class from among the candidates on the basis of the counted appearance numbers.

6. The method according to claim 4, wherein the error information includes a timestamp, a time period between times indicated by timestamps in the error information of the pair of error components is less than a predetermined time.

7. The method according to claim 4, further comprising:

generating investigation information that specifies failure component type data on the basis of the component type data of the pair of error components, the relation information related to the pair of the error components, the error history information related to the pair of the error components and a history of previous investigations including component type data of a previous error component and a previous failure component causing a previous error of the previous error component,

the investigation information corresponding to the relation class including a pair of the first component type data and the second component type data.

8. The method according to claim 4, further comprising:

the estimating of the point includes selecting a component on the basis of an appearance frequency of each of the plurality of components in the investigation range.

9. A computer-readable, non-transitory medium storing a program for managing a network system including a plurality of components, the program causing a computer to execute a procedure, the procedure comprising:

obtaining component type data of each component of the plurality of components, component relation data including relation information indicating a pair of components related to each other in the network system, the relation information including a first component type data of a first component requesting a process of the component and a second component type data of a second component accepting a request of the process requested by the first component, and error history data indicating error information of the plurality of components,

extracting candidates of a relation class indicating a propagating direction of an error between components, based on the component type data, component relation data and the error history data,

determining the relation class among the candidates based on the first component type data and the second component type data,

determining an investigation range by predetermining a specified hop value and decrementing the hop value every time the relation class is tracked, and

estimating a point where a failure causing the error has occurred by tracking the relation class within the investigation range.

10. The computer-readable, non-transitory medium according to claim 9, wherein the procedure further comprises

counting, for each combination of the pair of component type data, an appearance number of the relation class candidates including the identical combination of the pair of component type data to the each combination;

wherein the determining the relation class includes determining the relation class from among the candidates on the basis of the counted appearance numbers.

11. The computer-readable, non-transitory medium according to claim 9, wherein the error information includes a timestamp, and a time period between times indicated by timestamps in the error information of the pair of error components is less than a predetermined time.

12. The computer-readable, non-transitory medium according to claim 9, wherein

the procedure further includes generating investigation information that specifies failure component type data on the basis of the component type data of the pair of error components, the relation information related to the pair of the error components, the error history information related to the pair of the error components and a history of previous investigations including component type data of a previous error component and a previous failure component causing a previous error of the previous error component,

the investigation information corresponding to the relation class including a pair of the first component type data and the second component type data.

13. The computer-readable, non-transitory medium according to claim 9, wherein

the estimating the point includes selecting a component on the basis of an appearance frequency of each of the plurality of components in the investigation range.