📝 add affinity section

Signed-off-by: vankichi <[email protected]>

docs/user-guides/capacity-plannig.md

@@ -23,7 +23,7 @@ For example, if you want to insert 1 million vectors with 900 dimensions and the
 (900 × 32 + 256 ) × 1,000,000 × 3 = 8,7168,000,000 (bit) = 10.896 (GB)
 ```
 
-It is just the minimum required RAM.
+It is just the minimum required RAM for indexing.
 Considering the margin of RAM capacity, the minimum RAM capacity should be less than 60% of the actual RAM capacity.
 Therefore, the actual minimum RAM capacity will be:
 
@@ -32,12 +32,10 @@ Therefore, the actual minimum RAM capacity will be:
 ```
 
 <div class="warn">
-Memory usage is not enough in the minimum required RAM.<BR>
-E.g., there is a noisy problem, high memory usage for createIndex, high traffic needs more memory, etc...
-There may be A noisy neighbor problem, High memory usage, 
+In the production usage, memory usage may be not enough in the minimum required RAM.<BR>
+E.g., there are a noisy problem, high memory usage for createIndex (indexing on memory), high traffic needs more memory, etc.
 </div>
 
-
 ## Kubernetes cluster view
 
 ### Pod priority & QoS
@@ -91,7 +89,7 @@ The below table shows the condition for each QoS.
 |    QoS     |   request CPU   | request Memory |    limit CPU    | request Memory  | Sup.                                |
 | :--------: | :-------------: | :------------: | :-------------: | :-------------: | :---------------------------------- |
 | Guaranteed |       :o:       |      :o:       |       :o:       |       :o:       | All settings are required.          |
-| Guaranteed | :o: (:warning:) | :o: (:warning) | :o: (:warning:) | :o: (:warning:) | One to three settings are required. |
+| Burstable  | :o: (:warning:) | :o: (:warning) | :o: (:warning:) | :o: (:warning:) | One to three settings are required. |
 | BestEffort |       :x:       |      :x:       |       :x:       |       :x:       | No setting is required.             |
 
 Vald requires many RAM resources because of on-memory indexing, so we highly recommend that you do not specify a limit, especially for the Vald Agent.
@@ -124,7 +122,47 @@ It is easy to change by editing the `values.yaml` file and applying it.
 Please take care of pod priority and QoS.
 </div>
 
-### Component view
+### Node & Pod affinity
+
+Kubernetes scheduler often applies Pods on Node based on resource availability.
+
+In production usage, other components sometimes work on the Kubernetes cluster where the Vald cluster runs.
+Depending on the situation, you may want to deploy to a different Node: e.g., when running a machine learning component that requires high memory on an independent Node.
+
+In this situation, we recommend you to set the affinity/anti-affinity configuration for each Vald component.
+It is easy to change by editing your `values.yaml`.
+
+```yaml
+# e.g. Agent's affinity settings
+...
+  agent:
+  ...
+    affinity:
+      nodeAffinity:
+        preferredDuringSchedulingIgnoredDuringExecution: []
+        requiredDuringSchedulingIgnoredDuringExecution:
+          nodeSelectorTerms: []
+      podAffinity:
+        preferredDuringSchedulingIgnoredDuringExecution: []
+        requiredDuringSchedulingIgnoredDuringExecution: []
+      podAntiAffinity:
+        preferredDuringSchedulingIgnoredDuringExecution:
+          - weight: 100
+            podAffinityTerm:
+              topologyKey: kubernetes.io/hostname
+              labelSelector:
+              matchExpressions:
+                - key: app
+                  operator: In
+                  values:
+                    - vald-agent-ngt
+        requiredDuringSchedulingIgnoredDuringExecution: []
+  ...
+```
+
+For more information about Kubernetes affinity, please refer to [here](https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#affinity-and-anti-affinity)
+
+## Component view
 
 Depending on the customization of each component for each user, there are some points to be aware of.