Couchbase on Kubernetes

The Couchbase on Kubernetes tutorial documents how to deploy Couchbase on Kubernetes. The following diagram provides a high level overview of the Couchbase on Kubernetes architecture.

The following components will be deployed to a Kubernetes cluster with minimal CPU and memory allocations to control cost:

Couchbase community-4.5.1

Features

The following features exist with this infrastructure:

Use of official Couchbase Docker Image
Clusterized Couchbase of 3 instances
Automatic Self-healing of failing Pods
Automatic Self-healing of failing Nodes
Couchbase connectivity from outside of the Kubernetes
No automatic Clustering: Clustering process is done manually at first startup.
No automatic Scaling: Scaling of new instances needs generation of new deployments and services.

Rationale

Couchbase can be installed either directly on the OS, or using Docker containers to prepare the cluster. There are existing solutions of clustering Couchbase using Docker containers, but they lack some of the features that containers can bring such as self-healing in case of pod or node failure.

This tutorial is designed so that the entire flow of clustering can be seen as step-by-step process.

The architecture is a static cluster of 3 instances of Couchbase, using the official docker image of Couchbase. The cluster is also designed so that it can be accessed by a service that resides outside of the Kubernetes cluster.

It is designed as a tutorial guideline, not for production usage.

Kubernetes can ease the deployment and management of Couchbase by leveraging some of Kubernetes advanced features including:

Advanced Scheduling: Affinity and Anti-affinity

In this tutorial we haven't used support for Stateful Applications or Dynamic Storage Provision, that could enable automatic Scaling:

Tutorial

Clean up

Run the clean-up bash script to remove all compute resources created by this tutorial:

bash clean-up

Credits