You are looking at the documentation of a prior release. To read the documentation of the latest release, please visit here.

New to KubeDB? Please start here.

Horizontal Scaling Overview

This guide will give you an overview of how KubeDB Ops Manager scales up/down the number of members of a MSSQLServer.

Before You Begin

How Horizontal Scaling Process Works

The following diagram shows how KubeDB Ops Manager used to scale up the number of members of a MSSQLServer cluster. Open the image in a new tab to see the enlarged version.

Horizontal scaling Flow
Fig: Horizontal scaling process of MSSQLServer

The horizontal scaling process consists of the following steps:

  1. At first, a user creates a MSSQLServer CR.

  2. KubeDB provisioner operator watches for the MSSQLServer CR.

  3. When it finds one, it creates a PetSet and related necessary stuff like secret, service, etc.

  4. Then, in order to scale the cluster up or down, the user creates a MSSQLServerOpsRequest CR with the desired number of replicas after scaling.

  5. KubeDB Ops Manager watches for MSSQLServerOpsRequest.

  6. When it finds one, it halts the MSSQLServer object so that the KubeDB provisioner operator doesn’t perform any operation on the MSSQLServer during the scaling process.

  7. Then KubeDB Ops Manager will add nodes in case of scale up or remove nodes in case of scale down.

  8. Then the KubeDB Ops Manager will scale the PetSet replicas to reach the expected number of replicas for the cluster.

  9. After successful scaling of the PetSet’s replica, the KubeDB Ops Manager updates the spec.replicas field of MSSQLServer object to reflect the updated cluster state.

  10. After successful scaling of the MSSQLServer replicas, the KubeDB Ops Manager resumes the MSSQLServer object so that the KubeDB provisioner operator can resume its usual operations.

In the next doc, we are going to show a step-by-step guide on scaling of a MSSQLServer cluster using Horizontal Scaling.