Browse Prior Art Database

Method and Apparatus for intelligent location-aware problem determination in a Service Component Architecture based distributed environment

IP.com Disclosure Number: IPCOM000200361D
Publication Date: 2010-Oct-08
Document File: 6 page(s) / 65K

Publishing Venue

The IP.com Prior Art Database

Abstract

In Service Component Architecture based envrionment, it's hard and time consuming to find the root failed components that has cross dependency on each other. Especially when there are clusters in distribute envrionment, the cluster topology makes the problem determination even harder, thus time consuming to find the exact pychical machine with the actual failed component. Usually locating the failed points in the actucal pychical machine is one of the most important work for resolving the problems in distribute envrionment. This publication provides a solution that using location correlate key for each component, together with the component depends on information,it can dynamically generate the invocation diagram with location information in runtime.With this solution, it will aid to identify the failure component not only in which component fails,but also in which psychical machine fails,which significantly eliminate unnecessary efforts and decrease the Problem Determination cycle in distribute envrionment. Especially in clusters envrionment, it can help detect the exact machine within cluster that the failed component located.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 40% of the total text.

Page 01 of 6

Method and Apparatus for intelligent location -

based distributed environment

1

1.

..Background

Background


In Service Component Architecture based envrionment , the software are typically composed of group ofloose -couple components which interact with each other through interface and messages . During runtime, the message will flow from one component to another component . If any error or exception happened during the procedure , the typical problem determination process willrequire the analysis of the execution log and the component interaction logic , then identify which component causes the error or exception , finally conclude why the error/expection is reported for the specific component .

In a simplified case such as sequential flow of components , in order to determine which component results in the issue , it can be done by going through the component flow and finding out the first component where the error /expection happened. In the case that the relationship and interaction among different components are more complicated ,usually the failure point we find it 's not the actual root failure point . It requires the understanding of the component interaction logic of the solution and advanced analysis skill just for simply point out the actual failure component .

What's more, it usually hard to find in which pychical machine the failed points located .It's time consuming to find the real failed machine which located the actual failed point , especially in clusters envrionment which is widely used in current distribute envrionment . Pls see figure 1 as example, it has component invocation chain as follows , and the failed point is at component F .

(This page contains 00 pictures or other non-text object)

1

-aware problem determination in a Service Component Architecture

aware problem determination in a Service Component Architecture



Page 02 of 6

Figure 1: Current Invocation Diagram with Failed Component Info
But the invocation flow occurs in the distribute envrionment , there are three clusters with some members located in different machine just as figure 2. How to find which is the exact pychical machine for the failed points ?

2



Page 03 of 6

(This page contains 00 pictures or other non-text object)

Figure 2: Clusters in Distribute Envrionment

3



Page 04 of 6

Pior art Comparision :


There are patents for problem determination in distribute systems using parent-child directed diagram,but it cannot represent the exactly location for failed points in clusters environment, such as patent[1] and [2]. Patent [3] generate dynamic diagram to help problem determination,but it also cannot locate the exact test machine in distributed environment .

It's important to determine the exact machine the failed point located in a Service Component Architecture environment, especially in the complex cluster environment. WebSphere Process Server provide Cross-Component Trace[4] to do debugging SCA components, but it's not location related and on...