Quality of Services (QoS)

QoS Classes

When Kubernetes creates a Pod it assigns one of these QoS classes to the Pod:

• Guaranteed

• Burstable

• BestEffort

Guaranteed

Pod中所有容器都必須統一設置limits，並且設置參數都一致，如果有一個容器要設置requests，那麽所有容器都要設置，並設置參數同limits一致，那麽這個pod的QoS就是Guaranteed級別。 註：如果一個容器只指明limit而未設定request，則request的值等於limit值。

For a Pod to be given a QoS class of Guaranteed:

• Every Container in the Pod must have a memory limit and a memory request, and they must be the same.

• Every Container in the Pod must have a CPU limit and a CPU request, and they must be the same.

Burstable

pod中只要有一個容器的requests和limits的設置不相同，該pod的QoS即為Burstable。

A Pod is given a QoS class of Burstable if:

• The Pod does not meet the criteria for QoS class Guaranteed.

• At least one Container in the Pod has a memory or CPU request.

BestEffort

如果對於全部的resources來說requests與limits均未設置，該pod的QoS即為Best-Effort。

Practice

We use these 3 yaml files to produce 3 types of the QoS:

# Guraranteed
apiVersion: v1
kind: Pod
metadata:
  name: syrus-qos-demo
  namespace: syrusk8s
spec:
  containers:
  - name: qos-demo-ctr
    image: nginx
    resources:
      limits:
        memory: "200Mi"
        cpu: "700m"
      requests:
        memory: "200Mi"
        cpu: "700m"

# Burstable
apiVersion: v1
kind: Pod
metadata:
  name: syrus-qos-demo-2
  namespace: syrusk8s
spec:
  containers:
  - name: qos-demo-ctr-2
    image: nginx
    resources:
      limits:
        memory: "200Mi"
      requests:
        memory: "100Mi"

# BestEffort
apiVersion: v1
kind: Pod
metadata:
  name: syrus-qos-demo-3
  namespace: syrusk8s
spec:
  containers:
  - name: qos-demo-ctr-3
    image: nginx

Create pod:

kubectl apply -f qos.yaml --namespace=syrusk8s

View detailed information about the Pod:

kubectl get pod syrus-qos-demo --namespace=syrusk8s --output=yaml

Check the output:

spec:
  containers:
    ...
    name: qos-demo-ctr
    resources:
      requests:
        memory: 200Mi
    ...
  qosClass: Guaranteed/Burstable/BestEffort (depends on what yaml you used)

QoS優先級

3種QoS優先級從有低到高（從左向右）： Best-Effort pods -> Burstable pods -> Guaranteed pods

kill order: BestEffort-> Burstable -> Guaranteed

QoS pods被kill掉場景與順序

Best-Effort 類型的pods：系統用完了全部內存時，該類型pods會最先被kill掉。

Burstable類型pods：系統用完了全部內存，且沒有Best-Effort container可以被kill時，該類型pods會被kill掉。

Guaranteed pods：系統用完了全部內存、且沒有Burstable與Best-Effort container可以被kill，該類型的pods會被kill掉。 註：如果pod進程因使用超過預先設定的limites而非Node資源緊張情況，系統傾向於在其原所在的機器上重啟該container或本機或其他重新創建一個pod。

使用建議

如果資源充足，可將QoS pods類型均設置為Guaranteed。用計算資源換業務性能和穩定性，減少排查問題時間和成本。 如果想更好的提高資源利用率，業務服務可以設置為Guaranteed，而其他服務根據重要程度可分別設置為Burstable或Best-Effort，例如filebeat。