New to KubeDB? Please start here.
Druid Horizontal Scaling
This guide will give an overview on how KubeDB Ops-manager operator scales up or down Druid
cluster replicas of various component such as Coordinators, Overlords, Historicals, MiddleManager, Brokers and Routers.
Before You Begin
- You should be familiar with the following
KubeDB
concepts:
How Horizontal Scaling Process Works
The following diagram shows how KubeDB Ops-manager operator scales up or down Druid
database components. Open the image in a new tab to see the enlarged version.
The Horizontal scaling process consists of the following steps:
At first, a user creates a
Druid
Custom Resource (CR).KubeDB
Provisioner operator watches theDruid
CR.When the operator finds a
Druid
CR, it creates required number ofPetSets
and related necessary stuff like secrets, services, etc.Then, in order to scale the various components (i.e. Coordinators, Overlords, Historicals, MiddleManagers, Brokers, Routers) of the
Druid
cluster, the user creates aDruidOpsRequest
CR with desired information.KubeDB
Ops-manager operator watches theDruidOpsRequest
CR.When it finds a
DruidOpsRequest
CR, it halts theDruid
object which is referred from theDruidOpsRequest
. So, theKubeDB
Provisioner operator doesn’t perform any operations on theDruid
object during the horizontal scaling process.Then the
KubeDB
Ops-manager operator will scale the related PetSet Pods to reach the expected number of replicas defined in theDruidOpsRequest
CR.After the successfully scaling the replicas of the related PetSet Pods, the
KubeDB
Ops-manager operator updates the number of replicas in theDruid
object to reflect the updated state.After the successful scaling of the
Druid
replicas, theKubeDB
Ops-manager operator resumes theDruid
object so that theKubeDB
Provisioner operator resumes its usual operations.
In the next docs, we are going to show a step-by-step guide on horizontal scaling of Druid cluster using DruidOpsRequest
CRD.