Integrate AWS Cloud Map with Istio

This tutorial describes how to integrate AWS Cloud Map with your Istio service mesh.

Overview

AWS Cloud Map is a managed service registry provided by Amazon Web Services(AWS). If your applications in Istio mesh require accessing to an external service registered in Cloud Map, then you might want to utilize the endpoint information in Cloud Map. By creating ServiceEntry resources holding the endpoint information in Cloud Map, we can finely control and observe egresses to the service. However, Istio does not provide the functionality to automatically sync ServiceEntries with the corresponding records in Cloud Map. That’s where Istio Cloud Map Operator comes into play.

Istio Cloud Map Operator is designed for syncing Cloud Map data into Istio by pushing ServiceEntry to the Kube API server. It periodically checks the Cloud Map resources in AWS, and if there’s any update in the information, then it creates/updates a ServiceEntry resource in the k8s cluster.

Prerequisites

Before proceeding, make sure you have a Kubernetes cluster with Istio installed.

You can follow the prerequisites for instructions on how to install and setup Istio.

Deploying Istio Cloud Map Operator

To get started we need to download and deploy Istio Cloud Map Operator:

  1. Download the manifests located here.

  2. Create an AWS IAM identity with read access to AWS Cloud Map for the operator.

  3. Modify the YAML file aws-config.yaml as follows:

    1. Set the access key used by the operator by updating the values in the secret.
    apiVersion: v1
    kind: Secret
    metadata:
      name: aws-creds
    type: Opaque
    data:
      access-key-id: <base64-encoded-IAM-access-key-id> # EDIT ME
      secret-access-key: <base64-encoded-IAM-secret-access-key> # EDIT ME
    
    1. Set the target AWS region.
    apiVersion: v1
    kind: ConfigMap
    metadata:
      name: aws-config
    data:
      aws-region: us-west-2 # EDIT ME
    
  4. Apply the modified manifest using Kubernetes CLI.

Verify the deployment

Assuming that you have the following data in you AWS Cloud Map,

# list the service in Cloud Map
$ aws servicediscovery list-services | jq '.Services[] | "Name: \(.Name), Id: \(.Id)"'
"Name: getistio-external-service, Id: srv-ou6hvfmjpls2lev6"

# check namespace of your service
$ aws servicediscovery get-namespace --id $(aws servicediscovery get-service --id srv-ou6hvfmjpls2lev6 | jq -r '.Service.NamespaceId') | jq '.Namespace.Name'
"my-namespace"

# list endpoints
$ aws servicediscovery list-instances --service-id srv-ou6hvfmjpls2lev6 | jq '.Instances[] | .Attributes'
{
  "AWS_INSTANCE_IPV4": "52.192.72.89"
}

Then, you can verify the Kubernetes deployment by checking that a ServiceEntry is created, and it contains exactly the same endpoint information as in the AWS Cloud Map.

You can run the following command to get the YAML representation of the resource:

kubectl get serviceentries.networking.istio.io cloudmap-getistio-external-service.my-namespace -o yaml

Here’s how the output from the command above should look like:

apiVersion: networking.istio.io/v1beta1
kind: ServiceEntry
metadata:
spec:
  addresses:
  - 52.192.72.89
  endpoints:
  - address: 52.192.72.89
    ports:
      http: 80
      https: 443
  hosts:
  - getistio-external-service.my-namespace
  ports:
  - name: http
    number: 80
    protocol: HTTP
  - name: https
    number: 443
    protocol: HTTPS
  resolution: STATIC

Note that the host name getistio-external-service.my-namespace is in the follow format: ${Cloud Map's service name}.${service namespace}.