High Availability

Configuring your Zetaris Platform to run in a Highly Available (HA) manner, can be achieved by deploying an additional controller node to operate as a hot/warm standby, waiting activation.

Your Zetaris cluster is deployed in a master-slave configuration, where you master, also known as the cluster controller, coordinates queries on run-time between slaves, also known as workers.

Standard cluster

Setting up for HA

Your Zetaris HA environment will require use of Apache Zookeeper, to coordinate the failover, in addition to a hot/warm standby controller node.

Your cluster will effectively channel queries through the primary controller node, and in case of a failure in the primary controller node, Zookeeper will trigger the standby controller to activate as the primary.

Cluster with High Availability

Setting up for HA in the cloud

When deploying Zetaris in a cloud based infrastructure, like Azure, Amazon Web Services, or Google Cloud Platform, we recommend setting up your shared metastore, so that all nodes have a common repository, which are often provided by the vendor with built in HA.

High availability with shared metastore

Steps for deploying High Availability

The specific steps for implementing the coordination of Apache Zookeeper your master nodes, is slightly different depending on the OS we have installed on.