[Scalability] Enable Horizontal Pod Autoscaler (HPA) for Alcor deployment

kubernetes/services/api-gateway.yaml

@@ -82,6 +82,12 @@ spec:
         - image: api_gateway:v1.0
           name: apimanager-web
           imagePullPolicy: IfNotPresent
+          # The value is determined according to the actual situation, the following value is for the deployment on AWS
+          resources:
+            limits:
+              cpu: 500m
+            requests:
+              cpu: 200m
           command: ["java", "-jar", "/app/AlcorApiGateway-0.1.0.jar", "--spring.config.location=/etc/api/application.properties"]
           ports:
             - containerPort: 8080

kubernetes/services/api-gateway_hpa.yaml

@@ -0,0 +1,79 @@
+apiVersion: autoscaling/v2beta2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: apimanager-hpa
+spec:
+  scaleTargetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: apimanager
+  # if minReplicas is less than Deployment replicas value, it may cause scale down
+  minReplicas: 5
+  maxReplicas: 10
+  behavior:
+    scaleDown:
+      # Indicates that the stability window considers the expected state of the past (here within 300 sec) to prevent expansion and contraction
+      stabilizationWindowSeconds: 300
+      policies:
+        - type: Percent
+          value: 100
+          periodSeconds: 15
+      # The autoscaler will choose the strategy that affects the minimum number of Pods
+      selectPolicy: Min
+    scaleUp:
+      stabilizationWindowSeconds: 0
+      policies:
+        - type: Percent
+          value: 100
+          periodSeconds: 15
+        - type: Pods
+          value: 4
+          periodSeconds: 15
+      selectPolicy: Max
+  metrics:
+#   Set the average usage rate of the cpu, scale up if it exceeds 50
+    - type: Resource
+      resource:
+        name: cpu
+        target:
+          type: Utilization
+          averageUtilization: 50
+#    The average data volume per second reaches 1000 for scaling up
+#    - type: Pods
+#      pods:
+#        metric:
+#          name: packets-per-second
+#        target:
+#          type: AverageValue
+#          averageValue: 1k
+
+#    The value is derived from Ingress "main-route", scale up when the number of requests per second in Ingress reaches 2000
+#    - type: Object
+#      object:
+#        metric:
+#          name: requests-per-second
+#        describedObject:
+#          apiVersion: networking.k8s.io/v1beta1
+#          kind: Ingress
+#          name: main-route
+#        target:
+#          kind: Value
+#          value: 10k
+
+# HPA's current status data
+status:
+  currentReplicas: 1
+  desiredReplicas: 1
+  conditions:
+    - status: "True"
+      type: "True"
+#    - type: Object
+#      object:
+#        metric:
+#          name: requests-per-second
+#        describedObject:
+#          apiVersion: networking.k8s.io/v1beta1
+#          kind: Ingress
+#          name: main-route
+#        current:
+#          value: 10k

kubernetes/services/deleteYaml.sh

@@ -51,58 +51,72 @@ if [[ "$OSTYPE" == "linux-gnu" ]]; then
   kubectl delete configmap sg-configmap
   kubectl delete services sgmanager-service
   kubectl delete deployment sgmanager
+  kubectl delete hpa sgmanager-hpa
 
   kubectl delete configmap vpc-configmap
   kubectl delete services vpcmanager-service
   kubectl delete deployment vpcmanager
+  kubectl delete hpa vpcmanager-hpa
 
   kubectl delete configmap subnet-configmap
   kubectl delete services subnetmanager-service
   kubectl delete deployment subnetmanager
+  kubectl delete hpa subnetmanager-hpa
 
   kubectl delete configmap route-configmap
   kubectl delete services routemanager-service
   kubectl delete deployment routemanager
+  kubectl delete hpa routemanager-hpa
 
   kubectl delete configmap mac-configmap
   kubectl delete services macmanager-service
   kubectl delete deployment macmanager
+  kubectl delete hpa macmanager-hpa
 
   kubectl delete configmap ip-configmap
   kubectl delete services ipmanager-service
   kubectl delete deployment ipmanager
+  kubectl delete hpa ipmanager-hpa
 
   kubectl delete configmap port-configmap
   kubectl delete services portmanager-service
   kubectl delete deployment portmanager
+  kubectl delete hpa portmanager-hpa
 
   kubectl delete configmap quota-configmap
   kubectl delete services quotamanager-service
   kubectl delete deployment quotamanager
+  kubectl delete hpa quotamanager-hpa
 
   kubectl delete configmap node-configmap
   kubectl delete services nodemanager-service
   kubectl delete deployment nodemanager
+  kubectl delete hpa nodemanager-hpa
 
   kubectl delete configmap api-configmap
   kubectl delete services apimanager-service
   kubectl delete deployment apimanager
+  kubectl delete hpa apimanager-hpa
 
   kubectl delete configmap dpm-configmap
   kubectl delete services dataplanemanager-service
   kubectl delete deployment dataplanemanager
+  kubectl delete hpa dataplanemanager-hpa
 
   kubectl delete configmap eip-configmap
   kubectl delete services eipmanager-service
   kubectl delete deployment eipmanager
+  kubectl delete hpa eipmanager-hpa
 
   kubectl delete configmap ncm-configmap
   kubectl delete services netwconfigmanager-service
   kubectl delete deployment netwconfigmanager
+  kubectl delete hpa netwconfigmanager-hpa
 
   kubectl delete configmap gateway-configmap
   kubectl delete services gatewaymanager-service
   kubectl delete deployment gatewaymanager
+  kubectl delete hpa gatewaymanager-hpa
 
 
 elif [[ "$OSTYPE" == "darwin"* ]]; then

kubernetes/services/dpm_manager.yaml

@@ -80,6 +80,12 @@ spec:
         #- image: fwnetworking/controller:dpm_manager-v0.3.0
           name: dataplanemanager-web
           imagePullPolicy: IfNotPresent
+          # The value is determined according to the actual situation, the following value is for the deployment on AWS
+          resources:
+            limits:
+              cpu: 500m
+            requests:
+              cpu: 200m
           command: ["java", "-jar", "/app/AlcorDataPlaneManager-0.1.0.jar", "--spring.config.location=/etc/dpm/application.properties"]
           ports:
             - containerPort: 8080

kubernetes/services/dpm_manager_hpa.yaml

@@ -0,0 +1,79 @@
+apiVersion: autoscaling/v2beta2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: dataplanemanager-hpa
+spec:
+  scaleTargetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: dataplanemanager
+  # if minReplicas is less than Deployment replicas value, it may cause scale down
+  minReplicas: 5
+  maxReplicas: 10
+  behavior:
+    scaleDown:
+      # Indicates that the stability window considers the expected state of the past (here within 300 sec) to prevent expansion and contraction
+      stabilizationWindowSeconds: 300
+      policies:
+        - type: Percent
+          value: 100
+          periodSeconds: 15
+      # The autoscaler will choose the strategy that affects the minimum number of Pods
+      selectPolicy: Min
+    scaleUp:
+      stabilizationWindowSeconds: 0
+      policies:
+        - type: Percent
+          value: 100
+          periodSeconds: 15
+        - type: Pods
+          value: 4
+          periodSeconds: 15
+      selectPolicy: Max
+  metrics:
+#   Set the average usage rate of the cpu, scale up if it exceeds 50
+    - type: Resource
+      resource:
+        name: cpu
+        target:
+          type: Utilization
+          averageUtilization: 50
+#    The average data volume per second reaches 1000 for scaling up
+#    - type: Pods
+#      pods:
+#        metric:
+#          name: packets-per-second
+#        target:
+#          type: AverageValue
+#          averageValue: 1k
+
+#    The value is derived from Ingress "main-route", scale up when the number of requests per second in Ingress reaches 2000
+#    - type: Object
+#      object:
+#        metric:
+#          name: requests-per-second
+#        describedObject:
+#          apiVersion: networking.k8s.io/v1beta1
+#          kind: Ingress
+#          name: main-route
+#        target:
+#          kind: Value
+#          value: 10k
+
+# HPA's current status data
+status:
+  currentReplicas: 1
+  desiredReplicas: 1
+  conditions:
+    - status: "True"
+      type: "True"
+#    - type: Object
+#      object:
+#        metric:
+#          name: requests-per-second
+#        describedObject:
+#          apiVersion: networking.k8s.io/v1beta1
+#          kind: Ingress
+#          name: main-route
+#        current:
+#          value: 10k

kubernetes/services/elastic_ip_manager.yaml

@@ -47,6 +47,12 @@ spec:
         #- image: fwnetworking/controller:mac_manager-v0.3.0
           name: eipmanager-web
           imagePullPolicy: IfNotPresent
+          # The value is determined according to the actual situation, the following value is for the deployment on AWS
+          resources:
+            limits:
+              cpu: 500m
+            requests:
+              cpu: 200m
           command: ["java", "-jar", "/app/AlcorElasticIpManager-0.1.0.jar", "--spring.config.location=/etc/eip/application.properties"]
           ports:
             - containerPort: 8080

kubernetes/services/elastic_ip_manager_hpa.yaml

@@ -0,0 +1,79 @@
+apiVersion: autoscaling/v2beta2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: eipmanager-hpa
+spec:
+  scaleTargetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: eipmanager
+  # if minReplicas is less than Deployment replicas value, it may cause scale down
+  minReplicas: 5
+  maxReplicas: 10
+  behavior:
+    scaleDown:
+      # Indicates that the stability window considers the expected state of the past (here within 300 sec) to prevent expansion and contraction
+      stabilizationWindowSeconds: 300
+      policies:
+        - type: Percent
+          value: 100
+          periodSeconds: 15
+      # The autoscaler will choose the strategy that affects the minimum number of Pods
+      selectPolicy: Min
+    scaleUp:
+      stabilizationWindowSeconds: 0
+      policies:
+        - type: Percent
+          value: 100
+          periodSeconds: 15
+        - type: Pods
+          value: 4
+          periodSeconds: 15
+      selectPolicy: Max
+  metrics:
+#   Set the average usage rate of the cpu, scale up if it exceeds 50
+    - type: Resource
+      resource:
+        name: cpu
+        target:
+          type: Utilization
+          averageUtilization: 50
+#    The average data volume per second reaches 1000 for scaling up
+#    - type: Pods
+#      pods:
+#        metric:
+#          name: packets-per-second
+#        target:
+#          type: AverageValue
+#          averageValue: 1k
+
+#    The value is derived from Ingress "main-route", scale up when the number of requests per second in Ingress reaches 2000
+#    - type: Object
+#      object:
+#        metric:
+#          name: requests-per-second
+#        describedObject:
+#          apiVersion: networking.k8s.io/v1beta1
+#          kind: Ingress
+#          name: main-route
+#        target:
+#          kind: Value
+#          value: 10k
+
+# HPA's current status data
+status:
+  currentReplicas: 1
+  desiredReplicas: 1
+  conditions:
+    - status: "True"
+      type: "True"
+#    - type: Object
+#      object:
+#        metric:
+#          name: requests-per-second
+#        describedObject:
+#          apiVersion: networking.k8s.io/v1beta1
+#          kind: Ingress
+#          name: main-route
+#        current:
+#          value: 10k

kubernetes/services/gateway_manager.yaml

@@ -55,6 +55,12 @@ spec:
         #- image: fwnetworking/controller:gateway_manager-v0.3.0
           name: gatewaymanager-web
           imagePullPolicy: IfNotPresent
+          # The value is determined according to the actual situation, the following value is for the deployment on AWS
+          resources:
+            limits:
+              cpu: 500m
+            requests:
+              cpu: 200m
           command: ["java", "-jar", "/app/AlcorGatewayManager-0.1.0.jar", "--spring.config.location=/etc/app/application.properties"]
           ports:
             - containerPort: 8080