Are the benefits of containerization limited to ease of integration, testing, deployment and operations? David Grizzanti explains how his team at Comcast – which develops a highly performant, distributed software system supporting millions of customers and deployed across multiple data centers – moved large-scale, multi-data-center services from an architecture deployed on virtual machines supported by separate development and operations teams to one based on containers with Apache Mesos operated by a single DevOps team, sharing how Comcast overcame multiple challenges – some that were anticipated and many that were not.
David discusses the team’s journey from excited experimentation through the valleys of frustration and ultimately to a deeper appreciation for containerization and explains how they were able to get everything working; and optimize the architecture to support service discovery and multitenancy in some interesting ways.
Topics include: - Some of the basics: Environments, tools, choices - War games and surprises: IP addresses, logging, metrics - Team organization: Taking responsibility for development, integration, and operations of the platform - Application errors under load - Debugging performance - Scaling the application up - Supporting multi-tenancy - Simplifying the application’s core algorithm - Achieving superior performance
David Grizzanti is a principal software engineer at Comcast, where he oversees the development of multitenant software platforms that support tens of millions of customers across North America. David has more than 10 years of software experience. Prior to joining Comcast, he participated in three IaaS platform builds in his time at Sungard Availability Services. His general areas of interests include software architecture, DevOps/QA, and engineering leadership.