Lan Xia & Shelley Lambert
Lan is a team lead in Functional Verification Team in IBM Runtime Technologies. As a software developer for over 10 years, she has extensive experience in software development, web development, and test management. She is a committer on the open source project - Eclipse OpenJ9. And she is also an active contributor for Eclipse OpenJ9, AdoptOpenJDK and Eclipse OMR project. She creates and provides support for test frameworks, test infrastructure, CI test pipelines, various testing, and node.js based test result summary service. She is self-motivated and constantly experimenting with new technologies and techniques to assure better testing. Lan loves traveling and she really enjoys trying different authentic food on her travels.

  Shelley is a Test Lead for IBM Runtime Technologies team. She and her team test open and freely available JDK implementations and have delivered the test strategy, test code base, and test frameworks into the Eclipse OMR, Eclipse OpenJ9 and AdoptOpenJDK projects. She serves on the Technical Steering Committee at AdoptOpenJDK. She is a committer at OpenJ9 and AdoptOpenJDK and draws stories and lessons from her experiences in the open projects where she is most active. Shelley balances her career in software verification by teaching yoga and establishing and maintaining food forests within the city of Ottawa, Ontario, Canada.

 
Speech title: Dealing with Verification Data Overload

Massive amounts of test logs and console output are generated each day at the AdoptOpenJDK and Eclipse OpenJ9 projects because of the huge number of tests multiplied by the number of versions, platforms and implementations tested. This ‘noise’ from daily, personal, and pull request builds requires processing in order to consume and make sense of it (more than 6G of test output is produced daily from Eclipse OpenJ9 test jobs alone, much more from AdoptOpenJDK). In some cases, we can instrument the test code to be less verbose. In all cases, we must take this raw data and refine it to more easily and effectively understand what next action to take. This presentation covers our approaches to handling the copious amounts of verification data, in order to make it meaningful and to give guidance on next actions. From a quality assurance perspective, there is value in results summary and aggregation. We need best practices in the application of data visualization techniques, filtering, and categorization. We need to continuously evolve and improve, employing relevant technologies such as deep learning. Ultimately, the goal of our data refinery efforts needs to be to display the results efficiently, allow users to quickly narrow down the problems and better monitor the farm to support the complex continuous delivery pipelines. Who is it for? People who are interested in knowing how to develop practical solutions for processing a large amount of verification data. During this talk you will hear about: What are the practices in the application of data visualization techniques, filtering, and categorization that worked for us? What are the benefits and challenges of trying to apply deep learning to refine our data? What are we plan on doing for getting better data refinery?  
Back to main page