New to KubeDB? Please start here.

PgBouncer Compute Resource Autoscaling

This guide will give an overview on how KubeDB Autoscaler operator autoscales the database compute resources i.e. cpu and memory using pgbouncerautoscaler crd.

Before You Begin

• You should be familiar with the following KubeDB concepts:
  • PgBouncer
  • PgBouncerAutoscaler
  • PgBouncerOpsRequest

How Compute Autoscaling Works

The following diagram shows how KubeDB Autoscaler operator autoscales the resources of PgBouncer. Open the image in a new tab to see the enlarged version.

  

The Auto Scaling process consists of the following steps:

1. At first, a user creates a PgBouncer Custom Resource Object (CRO).

2. KubeDB Provisioner operator watches the PgBouncer CRO.

3. When the operator finds a PgBouncer CRO, it creates PetSet and related necessary stuff like secrets, services, etc.

4. Then, in order to set up autoscaling of PgBouncer, the user creates a PgBouncerAutoscaler CRO with desired configuration.

5. KubeDB Autoscaler operator watches the PgBouncerAutoscaler CRO.

6. KubeDB Autoscaler operator generates recommendation using the modified version of kubernetes official recommender for different components of the database, as specified in the PgBouncerAutoscaler CRO.

7. If the generated recommendation doesn’t match the current resources of the database, then KubeDB Autoscaler operator creates a PgBouncerOpsRequest CRO to scale the pgbouncer to match the recommendation generated.

8. KubeDB Ops-manager operator watches the PgBouncerOpsRequest CRO.

9. Then the KubeDB Ops-manager operator will scale the pgbouncer vertically as specified on the PgBouncerOpsRequest CRO.

In the next docs, we are going to show a step-by-step guide on Autoscaling of PgBouncer using PgBouncerAutoscaler CRD.