Self-Healing Metadata Based on Application Feedback
Publication Date: 2011-Aug-11
The IP.com Prior Art Database
A centralized metadata and business logic repository allows us to make data updates to several tools simultaneously. Described is a system that allows the issues to be triaged and fixed in the end-user tools instantaneously.
Page 01 of 3
This invention describes a centralized repository capable of holding and organizing metadata and configuration information, such as specialized business logic, for a variety of different tools. This area handles several different defect-tracking systems: IBM* Configuration Management Version Control (CMVC), Rational ClearQuest*, Rational Team Concert (RTC), to name a few, and they each use similar metadata and logic: releases can span tools and customizations such as not letting anyone but the team lead close a defect, etc. What is needed is a way to organize this overlapping data and reduce the amount of time spent administrating it. This repository allows making changes in one place and have those changes automatically shipped to one or more tools or same-tool repositories without administrators needing to update each tool separately. It saves a lot of time, and because the data is organized and automated, one does not forget to update corresponding information in a second tool once the first tool has been updated.
There is a problem that arises any time you automatically make changes to a tool's metadata or business logic: sometimes it doesn't work (this is why you have extensive test cycles for new code). However, that requires manpower and time when what one would really like to do is to be able to make a change in this centralized repository, ship it to multiple tools, and have the kinks work themselves out. This invention pays attention to error messages caused by metadata and configuration updates, and leverages the centralized nature of the single metadata and business logic repository to make a series of guesses on what would fix the problem and updates the tools' data automatically. It may go through several iterations before a fix is found or an administrator needs to be notified, but this can be done relatively quickly while a user is manipulating the system. Benefits would be:
1. Saves administrator time so they can be working on other things - features - instead of debugging data problems.
2. The concept could be extended to any application, not just tools like Rational ClearQuest and RTC. What developer would not want simple bugs to fix themselves?
Figure 1 below is a picture of how one could leverage the centralized metadata repository. Data updates here are shipped to several different tools. Once users exercise the new code or
new metadata and hit a problem, an error message can be captured and parsed by the Self-Healing Code, which has access to the data in the centralized repository and knowledge of
its organization. Error messages, information about the application, and the action a customer is performing can be captured either by customizing the applications themselves to send the messages back to the Self-Healing Code, or by having the Self-Healing code poll and collect this data in another manner. It then parses the information and...