Presentation: "Roadmap to Continuous Deployment"

Time: Wednesday 15:00 - 16:00

Location: Stanford

Abstract:

Modeling the complexities of large-scale online services is becoming increasingly infeasible. Continuous production deployment supplements traditional lab-based testing by providing software teams with insight into what really works. However, without properly architected systems, Continuous Deployment yields its own set of intractable problems.

'Roadmap to Continuous Deployment' outlines one large-scale online service provider's move from quarterly big-batch releases to Continuous Deployment. Once their system architecture evolved to support this new way of releasing software, four categories of technology patterns emerged: Service Isolation, Revision Rollback, Defect Detection, & Continuous Integration in Production.

Nathan Dye

 Nathan  Dye Nathan currently works for Microsoft Advertising. I joined Microsoft as part of the aQuantive acquisition two years ago. He has worked for aQuantive for the past 9 years, and has seen the R&D team grow from around 30 people to 300, and their systems scaled from 1,000 transactions per second to 250,000 transactions per second. It was during this time that they moved from big-batch releases to one aligned with continuous deployment.