In one of the earlier Kafka disaster recovery on Kubernetes with CSI post we discussed how brokers of an Apache Kafka clusters can be backed up and restored leveraging Kubernetes volume snapshots. While this solution provides a good enough disaster recovery option (and a super quick recovery), it doesn’t help when the entire Kubernetes cluster hosting the Kafka cluster is lost. In this post, we show you how to: back up a Kafka cluster to a remote Kafka cluster running on a separate Kubernetes cluster, and also how to recover the lost Kafka cluster with Supertubes.

Apache Kafka was designed with a heavy emphasis on fault-tolerance and high-availability in mind, and thus provides different methods of ensuring enterprise-grade resiliency such as: replication factor - which defines how many partition replicas of a topic should be kept, each one being stored on a different broker. A replication factor of N allows for the loss of N-1 brokers. The new brokers will replace the lost brokers until the number of remaining broker(s) reaches the replication factor, or the number of copies of the partitions of a topic.

Readers of this blog may remember a post we made in January about Bank-Vaults that touched on the topic of disaster recovery with multi datacenter replication. In that post we dicussed replication, mostly in the context of it being used as a form of hot backup. Today we’ll be exploring cold backups, another but equally important form of disaster recovery. Why we use Velero We use a toolset for Kubernetes disaster recovery called Velero.

Business continuity is a key requirement for our enterprise users, and it becomes exponentially more important in a cloud native environment. The Banzai Cloud Platform operates in such an environment, deploying and managing large scale container-based applications. It leverages best-of-breed cloud components, such as Kubernetes, to create a highly productive, yet flexible environment, and provides a safety net to backup and restore cluster and application states during the lifecycle of the cluster.