The KOST Stack


In MySQL, you have the LAMP stack. With MongoDB there is the MEAN stack. So what is the KOST Stack?

Because the TiDB Platform uses a component based architecture, we have some choices as to how we deploy it. KOST is an infrastructure pattern which consists of Kubernetes, Operator, Spark, and TiDB. We have already introduced the last two letters, so in this video I'm going to introduce Kubernetes and Operator:

  • Kubernetes is a container orchestration system that allows you to automate the deployment and management of containerized applications.
  • An Operator is designed to automate certain operations on a Kubernetes cluster. This pattern was first developed by the fine folks at CoreOS. It's often used to ensure fault tolerance. For TiDB, our operator implementation checks cluster health, ensures that Kubernetes is scheduling containers appropriately and also assists in software upgrades.

By being an orchestration system, it means that Kubernetes can help us setup all of the TiDB components in one step, removing the complexity that can come with distributed systems. The configuration system that works with Kubernetes to deploy many components is called Helm. PingCAP offers Helm charts for TiDB Operator, which is what we are going to be using for this course.

We will also be using Kubernetes+Operator to practice scaling our TiDB platform by increasing the number of TiDB and TiKV nodes.

Don't worry if some of this sounds like new terminology. We will be going through a tutorial to set everything up in just a second.