System and method for managing the performance of a computer system based on operational characteristics of the system components

• US 7,457,864 B2
• Filed: 11/27/2002
• Issued: 11/25/2008
• Est. Priority Date: 11/27/2002
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chambliss, David Darden, Pandey, Prashant, Chung, Leonard Chi Chiu, Wood, Eric Joseph, Belknap, Peter Clayton
• Primary Examiner(s): Zeender; F.
• Assistant Examiner(s): Refai; Ramsey
• Application Number:
  US10/306,853
• Publication Number:
  US 20040103181A1
• Time in Patent Office: 2,190 Days
• Field of Search: None
• US Class Current: 709/224
• CPC Class Codes: H04L 41/024 using relational databases for representation of network management data, e.g. managing via structured query language [SQL]
• CPC Class Codes: H04L 43/0817 by checking functioning

Abstract

A performance manager and method for managing the performance of a computer system based on a system model that includes measured entities representing the operational characteristics of the system components and relationships among the measured entities. The performance manager includes data producers for interacting with the interface agents of the components, an engine for exchanging information with the data producers and the system model, and an interaction model for determining relevant measured entities in the system model. The system model and interaction model are maintained in a repository where data might be accessed via an access interface. Incoming performance data is analyzed by an analyzer in the background to detect trends and relationships among the entities. An operator might review the relevant entities and apply controls to selected entities to manage the overall system performance as well as to resolve problems affecting the performance of the components in the system.

Claims

1. A method for managing performance of a computer system including a plurality of components each associated with a set of performance characteristics, the method comprising:

representing the state of the computer system as a system model, the system model including a plurality of measured entities and relationships among the measured entities, the measured entities representing the performance characteristics of the components;

providing the system model with performance, configuration and diagnostic information about the components;

determining a set of most relevant entities affecting the computer system performance based on the system model;

analyzing the performance information and detecting relationships among the measured entities, each relationship being represented as a data object in the system model and containing information about how metrics of an independent measured entity affect metrics of a dependent measured entity, wherein each measured entity corresponds to a plurality of metrics and controls, a forward coupling strength matrix is formed from values representative of a magnitude and a direction of a change in the dependent measured entity as a result of a change in the independent measured entity, a forward confidence matrix is formed from values representative of a probability that the corresponding value in the forward coupling strength matrix will affect a relationship between selected measured entities, the data object includes the result of a matrix multiplication of the forward coupling strength matrix and the forward confidence matrix, the plurality of controls affecting the operation of the components associated with said measured entity; and

changing, the controls associated with the set of most relevant measured entities to improve the performance of the computer system.

2. The method as recited in claim 1 further comprising presenting the set of most relevant entities to a system operator in an interaction session.

3. The method as recited in claim 1 further comprising diagnosing a performance problem in the computer system using a performance manager.

4. The method as recited in claim 1 further comprising selecting one of the measured entities as a focus entity;

and

the determining of the set of most relevant entities includes computing an entity priority function based on the focus entity and entity relationship.