Add documentation for defining HorizontalPodAutoscaler infinispan#2133

documentation/asciidoc/stories/assembly_auto_scaling.adoc

@@ -0,0 +1,27 @@
+ifdef::context[:parent-context: {context}]
+[id='auto-scaling']
+:context: scaling
+= Auto Scaling
+
+[role="_abstract"]
+Kubernetes includes the `HorizontalPodAutoscaler` which allows StatefulSets or Deployments to be automatically scaled up or
+down based upon specified metrics. The Infinispan CR exposes the `.status.scale` sub-resource, which enables `HorizontalPodAutoscaler`
+resources to target the Infinispan CR.
+
+Before defining a `HorizontalPodAutoscaler` configuration, consider the types of {brandname} caches that you define. Distributed
+and Replicated caches have very different scaling requirement, so defining a `HorizontalPodAutoscaler` for server's running
+a combination of these cache types may not be advantageous. For example, defining a `HorizontalPodAutoscaler` that scales
+when memory usage reaches a certain percentage will allow overall cache capacity to be increased when defining Distributed
+caches as cache entries are spread across pods, however it will not work with replicated cache as every pod hosts all cache
+entries. Conversely, configuring a `HorizontalPodAutoscaler` based upon CPU usage will be more beneficial for clusters
+with replicated cache as every pod contains all cache entries and so distributing read requests across additional nodes
+will allow a greater number of requests to be processed simultaneously.
+
+include::{topics}/proc_configuring_auto_scaling.adoc[leveloffset=+1]
+
+IMPORTANT: HorizontalPodAutoscaler should be removed when upgrading a {brandname} cluster, as the automatic scaling will
+cause the upgrade process to enter unexpected state, as the Operator needs to scale the cluster down to 0 pods.
+
+// Restore the parent context.
+ifdef::parent-context[:context: {parent-context}]
+ifndef::parent-context[:!context:]

documentation/asciidoc/titles/stories.adoc

@@ -13,6 +13,7 @@ include::{stories}/assembly_network_access.adoc[leveloffset=+1]
 include::{stories}/assembly_cross_site_replication.adoc[leveloffset=+1]
 include::{stories}/assembly_monitoring.adoc[leveloffset=+1]
 include::{stories}/assembly_anti_affinity.adoc[leveloffset=+1]
+include::{stories}/assembly_auto_scaling.adoc[leveloffset=+1]
 include::{stories}/assembly_cache_cr.adoc[leveloffset=+1]
 include::{stories}/assembly_batch_cr.adoc[leveloffset=+1]
 include::{stories}/assembly_backing_up_restoring.adoc[leveloffset=+1]

documentation/asciidoc/topics/proc_configuring_auto_scaling.adoc

@@ -0,0 +1,18 @@
+[id='configuring_auto-scaling-{context}']
+= Configuring HorizontalPodAutoscaler
+
+[role="_abstract"]
+Create a HorizontalPodAutoScaler resource that targets your Infinispan CR.
+
+.Procedure
+
+. Define a `HorizontalPodAutoscaler` resource in the same namespace as your `Infinispan` CR
++
+[source,options="nowrap",subs=attributes+]
+----
+include::yaml/horizontal_pod_autoscaler.yaml[]
+----
++
+<1> The name of your `Infinispan` CR
+
+NOTE: If using metric resource of type `cpu` or `memory`, you must configure request/limits for this resource in your `Infinispan` CR.

documentation/asciidoc/topics/yaml/horizontal_pod_autoscaler.yaml

@@ -0,0 +1,18 @@
+apiVersion: autoscaling/v2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: infinispan-auto
+spec:
+  scaleTargetRef:
+    apiVersion: infinispan.org/v1
+    kind: Infinispan
+    name: example # <1>
+  minReplicas: 1
+  maxReplicas: 10
+  metrics:
+    - type: Resource
+      resource:
+        name: cpu
+        target:
+          type: Utilization
+          averageUtilization: 50