Browse Prior Art Database

Method and Apparatus for Bottleneck Inference

IP.com Disclosure Number: IPCOM000016774D
Original Publication Date: 2003-Jul-15
Included in the Prior Art Database: 2003-Jul-15
Document File: 5 page(s) / 35K

Publishing Venue

IBM

Abstract

Disclosed are methods and apparatuses for automatically acquiring logs and configurations from each sub-system in a Web system to infer performance bottlenecks of the entire system using inference engines based on these acquisition data, and using visual analysis of these data for ensuring the correctness of inference engines. For keeping high availability and performance in the system such as a Web site, it is necessary to monitor and analyze it for grasping status to infer bottlenecks or their candidates. However, total amount of logs from each sub-system such as databases or application servers that consist of the focusing system is huge. Hence, it is difficult to manually analyze such logs and it is almost impossible to infer bottlenecks correctly from such huge logs. Disclosed methods and apparatuses get around these issues.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 25% of the total text.

Page 1 of 5

Method and Apparatus for Bottleneck Inference

  Disclosed are methods and apparatuses for automatically acquiring logs and configurations from each sub-system in a Web system to infer performance bottlenecks of the entire system using inference engines based on these acquisition data, and using visual analysis of these data for ensuring the correctness of inference engines. For obtaining high-availability and high-performance of a system, it is necessary for an administrator to observe and analyze logs for grasping the status and taking appropriate actions. However, total amount of logs from each sub-system such as databases or application servers that consist of the focusing system is huge. Hence, it is almost impossible to analyze correlation of these logs by hand to infer bottlenecks correctly. The reasons of these difficulties are considered as follows: The number of combination between logs needed in bottleneck analysis tends to increase. Hence, the complexity of such analysis also increases. These are issues to be solved in system monitoring methods and apparatuses for high availability and performance systems.

 Getting around the above described issues, this disclosed method and apparatus automatically acquire logs and configurations from each sub-system to infer performance bottlenecks using inference engines based on these acquisition data, visually analyze these data for ensuring the correctness of inference engines, and show advices or action examples for resolving bottlenecks in the system. This leads to avoidance of check failure of important information in huge amount logs and to reduction of workload for manual analysis and inference. At the same time, such combination of the inference part and the visual analysis part forms a complementary role for easy treatment and acquisition method to a new bottleneck problem that cannot be solved by the inference engine based on previous knowledge. That is, the method and apparatus automatically acquires the information of predefined or user selected log-items under a load-test environment to the Web-site or an actual load environment, analyzes visually these acquired information using various axis that users select depending on their intention, and automatically gives advices to users about bottleneck candidates for resolving them using inference agents that are based on the concept of JavaBeans or similar concept of such encapsulation methods. And the relationship between visual analysis and automatic reasoning method is complementary as follows: the result of visual analysis method has a reference role to the inference engine results. On the other hand, based on the rational comprehension through the visual analysis, the result of the automatic inference method is ensured. Hence, through combination of these complementary two methods, a new resolving knowledge and experiment to a new bottleneck example could be newly included in the inference agents as assets.

An example of sp...