System and method for monitoring and modeling system performance

• US 6,973,415 B1
• Filed: 08/09/2004
• Issued: 12/06/2005
• Est. Priority Date: 11/12/2003
• Status: Expired due to Fees

• Current Assignee: Sprint Communications Company LP (Deutsche Telekom AG)
• Original Assignee: Sprint Communications Company LP (Deutsche Telekom AG)
• Inventors: Saghier, Dean Lee, Washburn, Brian, Schroder, Scott
• Primary Examiner(s): Assouad, Patrick J.
• Application Number:
  US10/914,405
• Time in Patent Office: 484 Days
• Field of Search: 702/186, 702/179, 702/182, 703/21, 703/22, 717/104, 717/124, 717/127, 717/131, 717/134, 717/135, 709/223, 709/224, 709/225, 709/226
• US Class Current: 702/186
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• CPC Class Codes: G05B 23/0294 Optimizing process, e.g. process efficiency, product quality
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 10/0639 Performance analysis of employees; Performance analysis of enterprise or organisation operations

Abstract

The present invention provides a method for modeling the performance of a system comprising computer software operating on computer hardware. In accordance with the present invention, system performance is modeled by fitting non-linear curves to data points for system performance parameters, such as response time and throughput, as a function of load. Data points may be measured in testing may be measured through monitoring a system operating in a production environment. While a variety of non-linear curves may be used in accordance with the present invention, a logarithmic curve may be used to model system throughput and an exponential curve may be used to model system response time. By defining a relationship between throughput and response time a distance may be calculated between the curves, and this distance may be used to determine an optimal load. Additionally, a graph displaying both throughput and response time as a function of load may be displayed to a system operator in a graphical user interface to facilitate the evaluation of system performance.

Claims

1. A method for monitoring system performance, the method comprising:

defining system metrics for a system to be monitored;

measuring the defined system metrics at predetermined time intervals, the measuring comprising measuring a first subset of the defined system metrics at a first predetermined time interval and measuring a second subset of the defined system metrics at a second predetermined time interval;

modeling the defined system metrics over time; and

identifying correlations between defined system metrics.

2. The method for monitoring system performance of claim 1, further comprising:

determining the strength of any identified correlations between defined system metrics.

3. The method for monitoring system performance of claim 1, wherein identifying correlations between defined system metrics further comprises identifying non-linear relationships between defined system metrics.

4. At least one computer readable media having embodied thereon computer readable instructions for causing at least one computer to perform a method for monitoring system performance, the method comprising:

receiving definitions of system metrics for a system to be monitored;

measuring the defined system metrics at predetermined time intervals, the measuring comprising measuring a first subset of the defined system metrics at a first predetermined time interval and measuring a second subset of the defined system metrics at a second predetermined time interval;

modeling the defined system metrics over time; and

identifying correlations between defined system metrics.

5. The at least one computer readable media of claim 4, the method further comprising:

determining the strength of any identified correlations between defined system metrics.

6. The at least one computer readable media of claim 4, wherein identifying correlations between defined system metrics further comprises identifying non-linear relationships between defined system metrics.