istio-http traffic management (1)

reference

1. Define objectives and rules

istio service grid to service a further abstraction:
- Pod can be used to label specific service process group;
- You can define load balancing strategy services;
- TLS can be specified requirements for the service;
- You can set the size of the connection pool for the service.
istio, the different groups of the same back-end service is called 子集(Subset), is also often referred to 服务版本.
In istio, the recommendations set clear goals for each grid access rules, after the adoption of istio flow control, choose a clear subset to be accessed according to the rules specified in sub or centralized traffic policy, such rules are It referred to DestinationRule, the following examples:
```
cat flaskapp-destinationrule.yaml
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: flaskapp
spec:
  host: flaskapp.default.svc.cluster.local
  trafficPolicy:
    loadBalancer:
      simple: LEAST_CONN
  subsets:
  - name: v1
    labels:
      version: v1
    trafficPolicy:
      loadBalancer: ROUND_ROBIN
  - name: v2
    labels:
      version: v2
```
Note that local rules are as follows:
- host: the necessary fields, representing a Serviceresource, or an ServiceEntryexternal service definition.
  - Recommendation: To prevent the same name in different namespaces service, you can use the fully qualified name.
- trafficPolicy: traffic strategy, DestinationRuleand Subsetcan be defined, Subsetthe higher the level.
- subsets: The tag field is used to select subsets define different.

2. Define a default route

2.1 Definitions default route

Each service is recommended to create a default route, in the absence of a specific routing rules, use the default routing rules for access to the specified subset, a service in order to ensure the stability of behavior By default, the sample is as follows:

cat flaskapp-default-vs.yaml 
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:
  name: flaskapp
spec:
  hosts:
  - flaskapp.default.svc.cluster.local
  http:
  - route:
    - destination:
        host: flaskapp.default.svc.cluster.local
        subset: v1

VirtualService: hub during istio flow control, is responsible for screening and forwarding traffic.
- For the host name work, but hoststhe field is an array of content, it can work for multiple host names.
- It can serve a variety of traffic protocols, such as http, tcp, tlsand so on.
In httpthe next stage, i.e., the specific routing rules.
- It supports multiple routes.

2.2 verification

# 应用目标规则与默认路由
kubectl apply -f flaskapp-destinationrule.yaml
kubectl apply -f flaskapp-default-vs.yaml

# 验证，可同步观察 kiali & Jaeger 等
kubectl exec -it -c sleep $(kubectl get pod -l app=sleep,version=v1 -o jsonpath='{.items..metadata.name}') /bin/bash
bash-4.4# for i in `seq 100` ; do http --body http://flaskapp.default/env/version ; done

2.3 summary

In istio when deploying a business application, it is recommended:

Use appthe label indicates that the application identity;
Use versionthe label indicates that the application version;
Creating goals rule;
Create a default routing rules, default route was configured list should become necessary to deploy content services under the grid environment.

3. Traffic splitting and migration

3.1 Weight

cat flaskapp-default-vs.yaml 
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:
  name: flaskapp
spec:
  hosts:
  - flaskapp.default.svc.cluster.local
  http:
  - route:
    - destination:
        host: flaskapp.default.svc.cluster.local
        subset: v1
      weight: 70
    - destination:
        host: flaskapp.default.svc.cluster.local
        subset: v2
      weight: 30
      
# 应用
kubectl apply -f flaskapp-default-vs.yaml

# 验证，可同步观察 kiali & Jaeger 等
# 因是权重的原因，验证时样本量需要大一些
kubectl exec -it -c sleep $(kubectl get pod -l app=sleep,version=v1 -o jsonpath='{.items..metadata.name}') /bin/bash
bash-4.4# for i in `seq 100` ; do http --body http://flaskapp.default/env/version ; done | awk -F "v1" '{print NF-1}'

important point:

The right flow distribution is heavy, and the sum of the weights must be 100;
If you do not explicitly declared weights, the default value is 100.

3.2 Object get istio

kubectl get, kubectl api-resourcesLists the current cluster support all object types
Tag -> IST config