Presentation: TweetFighting Zombies: With Containers towards Fault-Tolerant Infrastructure

A fault-tolerant infrastructure is not only highly available but much more robust and simple to operate than traditional setups. When implemented systematically, it can be a game changer for operations cost and agility. One of the benchmarks there is Google's infrastructure. By combining container technologies like Docker and Mesos with distributed storage it is now finally possible to play the same game and build an infrastructure with the similar properties around fault-tolerance, scalability and flexibility.

In this talk we review the benefits of a scalable and fault-tolerant infrastructure build on containers and cluster-level infrastructure and investigate the problems that need to be solved. In particular data and state play a central role there, and zombies are one of the easier adversaries to defeat. We take a look at the puzzle pieces that are already available and sketch how they can be combined to build next generation infrastructures.