[ Forecasting Production Readiness - A Predictive Analytics Approach ]
Predictive analytics is an area of data mining that deals with extracting information from data and using it to predict trends and behavior patterns. In this paper, we explore the possibility of mining CI execution data to setup prediction models for the production readiness of CI builds.
Constructing such a model would require for us to look at specific behavioral patterns and epochs which are known causes of failures such as:
- Technical Stack Upgrades (epoch)
- Test failures occurring for more than once (pattern)
- Source code commits that challenge integration sanity. (pattern)
These factors will help us construct a “descriptive model” and setup a scoring system for the builds that will help us with the forecasting.
Being able to forecast when and which production builds are actually release worthy, gives us the following:
- Realistic look at timelines
- Multi-dimensional insights into the development/testing processes
- Quantifiable savings on the continuous delivery system
Raghavendra Sathyanarayana has over 19 years of experience in multiple disciplines including software testing,. He currently heads the Continuous Integration Build and Infrastructure team as Director, at Manhattan Associates.
Anil Satya is a Principal Software Engineer in the R&D organization at Manhattan Associates India. He has over 14 years of experience in the industry primarily in the Product Engineering space. He has been with Manhattan Associates from July 2003. In his role, he is responsible for Continuous Integration Build and Infrastructure for global R&D teams for Manhattan Suite of products. He has been instrumental in setting up the Automation Platforms and processes for both Product Development and Services. He has also worked on Analytics platform services for Continuous Testing and Delivery.
Prior to getting on the CI team, he has also worked in a Quality Assurance Engineer role and a 3-year specialty project for test architecture. He has played multiple roles which includes Testing, Development, and Build/Systems engineering mainly for the Supply Chain Management industry. He has presented an Analytics based paper earlier at Step-IN.