Kubernetes Meetup 18 10 2017 Johannes Scheuermann

Описание

Talk given at https://www.meetup.com/Rhein-Neckar-Kubernetes

"Data-Aware Scheduling in Kubernetes", by Johannes Scheuermann

In order to provide prompt results and efficiently deal with data-intensive workloads, Big Data applications execute their jobs on compute slots across large clusters. Also, for optimal performance, these applications should be as close as possible to the data they use. Data-aware scheduling is the way to achieve that optimization and can conveniently be set up using Kubernetes. We’ll present two different use cases: First, we’ll make use of how Big Data applications like Hadoop and Spark can use their native HDFS protocol for data-aware scheduling. Second, we’ll demonstrate an efficient way to write a data-aware scheduler for Kubernetes that satisfies not just your application’s requirements, but also keeps your admins happy. As a bonus, it’ll also allows us to run data-aware scheduling on applications other than Big Data.

About Johannes:
Johannes is an operations guy with a software development background. He's been working with Containers and Kubernetes for over years. He has a strong interest in modern infrastructures and building infrastructure software.