diff --git a/README.md b/README.md
index 3b4fac2..e0399c2 100644
--- a/README.md
+++ b/README.md
@@ -222,6 +222,16 @@ spec:
 
 <a id="helm">
 
+## PodDisruptionBudget
+
+If you want to ensure that more than 1 node won't be unavailable at a time, you can add PodDisruptionBudget manifests:
+```
+kubectl create -f es-master-pdb.yaml
+kubectl create -f es-data-pdb.yaml
+```
+
+Note: It's not recommended to grow this number because 1 node could be in maintenance, another fall down and you can't lose another node if the number of replicas is equal to two.
+
 ## Deploying with Helm
 
 [Helm](https://github.com/kubernetes/helm) charts for a basic (non-stateful) ElasticSearch deployment are maintained at https://github.com/clockworksoul/helm-elasticsearch. With Helm properly installed and configured, standing up a complete cluster is almost trivial:
diff --git a/es-data-pdb.yaml b/es-data-pdb.yaml
new file mode 100644
index 0000000..2ac8240
--- /dev/null
+++ b/es-data-pdb.yaml
@@ -0,0 +1,10 @@
+apiVersion: policy/v1beta1
+kind: PodDisruptionBudget
+metadata:
+  name: elasticsearch-data
+spec:
+  maxUnavailable: 1
+  selector:
+    matchLabels:
+      component: elasticsearch
+      role: data
diff --git a/es-master-pdb.yaml b/es-master-pdb.yaml
new file mode 100644
index 0000000..eff3d2a
--- /dev/null
+++ b/es-master-pdb.yaml
@@ -0,0 +1,10 @@
+apiVersion: policy/v1beta1
+kind: PodDisruptionBudget
+metadata:
+  name: elasticsearch-master
+spec:
+  maxUnavailable: 1
+  selector:
+    matchLabels:
+      component: elasticsearch
+      role: master