Tencent Cloud

Recent Pages

Locality Load Balancing

Last updated: 2023-12-26 11:36:12
Overview
This document describes how to deploy another set of website businesses in a cluster in another AZ to increase availability when the business expands. In particular, two sets of the same website businesses are available in two clusters at the same time. Locality load balancing is as shown below:
﻿
﻿
Directions
When the two sets of website businesses run normally, the ingress gateway will preferentially route traffic to the frontend service in the local region or AZ, even though the other cluster also has the frontend service. In addition, the frontend service will access the user, product, order, and cart services in the same AZ, and the order and cart services will access the stock service in the same AZ.
In Kubernetes, the Pod region is determined by the region and zone labels of the deployed node. The zone label is set for a workload in the YAML content of the demo application. First, deploy all the website services to the cluster (sub-cluster) in another AZ:
apiVersion: v1
kind: Namespace
metadata:
  name: base
  labels:
    istio.io/rev: 1-6-9
spec:
  finalizers:
    - kubernetes
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: frontend
  namespace: base
  labels:
    app: frontend
spec:
  replicas: 1
  selector:
    matchLabels:
      app: frontend
  template:
    metadata:
      labels:
        app: frontend
    spec:
      containers:
        - name: frontend
          image: ccr.ccs.tencentyun.com/chloeyhuang/demo:v202007101540
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 80
---
﻿
apiVersion: v1
kind: Service
metadata:
  name: frontend
  namespace: base
  labels:
    app: frontend
spec:
  ports:
    - port: 80
      name: http
  selector:
    app: frontend
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: product-v1
  namespace: base
  labels:
    app: product
    version: v1
spec:
  replicas: 1
  selector:
    matchLabels:
      app: product
      version: v1
  template:
    metadata:
      labels:
        app: product
        version: v1
    spec:
      containers:
        - name: product
          image: ccr.ccs.tencentyun.com/zhulei/testproduct1:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: product-v2
  namespace: base
  labels:
    app: product
    version: v2
spec:
  replicas: 1
  selector:
    matchLabels:
      app: product
      version: v2
  template:
    metadata:
      labels:
        app: product
        version: v2
    spec:
      containers:
        - name: product
          image: ccr.ccs.tencentyun.com/zhulei/testproduct2:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
---
﻿
apiVersion: v1
kind: Service
metadata:
  name: product
  namespace: base
  labels:
    app: product
spec:
  ports:
    - port: 7000
      name: http
  selector:
    app: product
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: user
  namespace: base
  labels:
    app: user
spec:
  replicas: 1
  selector:
    matchLabels:
      app: user
  template:
    metadata:
      labels:
        app: user
    spec:
      containers:
        - name: user
          image: ccr.ccs.tencentyun.com/zhulei/testuser:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
---
﻿
apiVersion: v1
kind: Service
metadata:
  name: user
  namespace: base
  labels:
    app: user
spec:
  ports:
    - port: 7000
      name: http
  selector:
    app: user
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: stock
  namespace: base
  labels:
    app: stock
spec:
  replicas: 1
  selector:
    matchLabels:
      app: stock
  template:
    metadata:
      labels:
        app: stock
    spec:
      containers:
        - name: stock
          image: ccr.ccs.tencentyun.com/zhulei/teststock:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
---
﻿
apiVersion: v1
kind: Service
metadata:
  name: stock
  namespace: base
  labels:
    app: stock
spec:
  ports:
    - port: 7000
      name: http
  selector:
    app: stock
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: cart
  namespace: base
  labels:
    app: cart
spec:
  replicas: 3
  selector:
    matchLabels:
      app: cart
  template:
    metadata:
      labels:
        app: cart
    spec:
      containers:
        - name: cart
          image: ccr.ccs.tencentyun.com/zhulei/testcart:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
              protocol: TCP
---
﻿
apiVersion: v1
kind: Service
metadata:
  name: cart
  namespace: base
  labels:
    app: cart
spec:
  ports:
    - name: http
      port: 7000
      protocol: TCP
  selector:
    app: cart
  type: ClusterIP
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: order-v1
  namespace: base
  labels:
    app: order
    version: v1
spec:
  replicas: 1
  selector:
    matchLabels:
      app: order
      version: v1
  template:
    metadata:
      labels:
        app: order
        version: v1
    spec:
      containers:
        - name: order
          image: ccr.ccs.tencentyun.com/zhulei/testorder1:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
              protocol: TCP
---
﻿
apiVersion: apps/v1
kind: Deployment
metadata:
  name: order-v2
  namespace: base
  labels:
    app: order
    version: v2
spec:
  replicas: 1
  selector:
    matchLabels:
      app: order
      version: v2
  template:
    metadata:
      labels:
        app: order
        version: v2
    spec:
      containers:
        - name: order
          image: ccr.ccs.tencentyun.com/zhulei/testorder2:v1
          imagePullPolicy: Always
          env:
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name
            - name: REGION
              value: "guangzhou-zoneB"
          ports:
            - containerPort: 7000
              protocol: TCP
---
﻿
apiVersion: v1
kind: Service
metadata:
  name: order
  namespace: base
  labels:
    app: order
spec:
  ports:
    - name: http
      port: 7000
      protocol: TCP
  selector:
    app: order
  type: ClusterIP
After the deployment, locality load balancing will not take effect if health check is not configured. In this case, services are called randomly in the two AZs, without adherence to the nearby access principle.
The order service calls the stock service in another AZ as shown below:
﻿
﻿
To enable locality load balancing for service access, you need to configure the health check feature for all the services by submitting the following YAML file to the primary cluster:
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: cart
  namespace: base
spec:
  host: cart
  trafficPolicy:
    loadBalancer:
      consistentHash:
        httpHeaderName: UserID
    outlierDetection:
      consecutiveErrors: 5
      interval: 10000ms
      baseEjectionTime: 30000ms
      maxEjectionPercent: 10
      minHealthPercent: 50
  exportTo:
    - '*'
﻿
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: frontend
  namespace: base
spec:
  host: frontend
  trafficPolicy:
    outlierDetection:
      consecutiveErrors: 5
      interval: 10000ms
      baseEjectionTime: 30000ms
      maxEjectionPercent: 10
      minHealthPercent: 50
  exportTo:
    - '*'
﻿
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: order
  namespace: base
spec:
  host: order
  trafficPolicy:
    outlierDetection:
      consecutiveErrors: 5
      interval: 10000ms
      baseEjectionTime: 30000ms
      maxEjectionPercent: 10
      minHealthPercent: 50
  subsets:
    - name: v1
      labels:
        version: v1
    - name: v2
      labels:
        version: v2
  exportTo:
    - '*'
﻿
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: product
  namespace: base
spec:
  host: product
  trafficPolicy:
    outlierDetection:
      consecutiveErrors: 5
      interval: 10000ms
      baseEjectionTime: 30000ms
      maxEjectionPercent: 10
      minHealthPercent: 50
  subsets:
    - name: v1
      labels:
        version: v1
    - name: v2
      labels:
        version: v2
﻿
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: stock
  namespace: base
spec:
  host: stock
  trafficPolicy:
    outlierDetection:
      consecutiveErrors: 5
      interval: 10000ms
      baseEjectionTime: 30000ms
      maxEjectionPercent: 10
      minHealthPercent: 50
  exportTo:
    - '*'
﻿
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: user
  namespace: base
spec:
  host: user
  trafficPolicy:
    outlierDetection:
      consecutiveErrors: 5
      interval: 10000ms
      baseEjectionTime: 30000ms
      maxEjectionPercent: 10
      minHealthPercent: 50
  exportTo:
    - '*'
After health check is configured, when a user accesses a website service, browses a product page, adds an item to the cart, or places an order through the edge gateway in the cluster of AZ A, the gateway will route the traffic to the frontend service in the same AZ, and the frontend service will also call the user, cart, order, and stock services in the same AZ. Similarly, when a user accesses a website service through the edge gateway of AZ B, the request will be routed to the frontend service in AZ B, and services in AZ B will be called. You can view the AZ information of the currently called service in the floating window in the bottom-left corner of the demo page. Locality load balancing is as shown below:
﻿
﻿