Deploying Envoy as an API Gateway for Microservices

An API Gateway is a façade that sits between the consumers and producers of an API. Cross-cutting functionality such as authentication, monitoring, and traffic management is implemented in your API Gateway so that your services can remain unaware of these details. In addition, when multiple services are responsible for different APIs (e.g., in a microservices architecture), an API Gateway hides this abstraction detail from the consumer.

There are dozens of different options for API Gateways, depending on your requirements. The Amazon API Gateway is a hosted Gateway that runs in Amazon. You could build your own on top of a Layer 7 proxy such as Traefik, NGINX, HAProxy, or Envoy. These all have their various strengths and weaknesses.

In general, though, you want to pick an API gateway that can accelerate your development workflow. Traditional API gateways focus on the challenges of API management, so using an API gateway that enables rapid development of services is essential.

Here at Datawire, we've been using Envoy for microservices. Envoy is interesting because, in addition to providing the reverse proxy semantics you need to implement an API Gateway, it also supports the features you need for distributed architectures (in fact, the Istio project builds on Envoy to provide a full-blown services mesh).

So let's take a closer look at deploying Envoy as a full-fledged, self-service API gateway. If you've been following along with our Envoy tutorial so far, we've done the following:

1. Create a Docker container based on the official Docker image
2. Crafted an Envoy configuration file
3. Deployed Envoy with the appropriate configuration file

Introducing Edge Stack API Gateway

This approach starts to get cumbersome as you add complexity to your deployment. For example, every configuration change requires editing a (complex!) configuration file, and redeploying Envoy. And, we've glossed over the operational aspects of keeping multiple Envoy instances running for scalability and availability.

We thought there would be an easier way, so we wrote Edge Stack. Here's what Edge Stack API Gateway does:

• Makes it easy to change and add to your Envoy configuration via Kubernetes annotations
• Adds the out-of-the-box configuration necessary for production Envoy, e.g., monitoring, health/liveness checks, and more
• Extends Envoy with traditional API Gateway functionality such as authentication
• Integrates with Istio, for organizations who need a full-blown service mesh

Edge Stack is a Kubernetes-native API Gateway. This means that it delegates all the hard parts of scaling and availability to Kubernetes. Want to upgrade Edge Stack with no downtime? No problem -- just use a Kubernetes rolling update.

Setting Up

We're going to assume that your basic infrastructure is set up enough that you have a Kubernetes cluster running in your cloud environment of choice. For now, we assume that:

• You have
  kubectl
  correctly talking to a Kubernetes cluster running in EC2 or GKE.
  • This is probably obvious, but it's tough to work with a Kubernetes cluster if you can't talk to it with
    kubectl
    .
• You have
  docker
  installed and working.
  • Since we'll be building Docker images, we need a working
    docker
    command.
• You have credentials to push Docker images to either Docker Hub or the Google Container Registry (gcr.io).

That last point is worth a little more discussion. To run something in Kubernetes, we have to be able to pull an Docker image from somewhere that the cluster can reach. When using Minikube, this is no problem, since Minikube runs its own Docker daemon: by definition, anything in the Minikube cluster can talk to that Docker daemon. However, things are different once GKE or EC2 come into play: they can't talk to a Docker daemon on your laptop without heroic measures, so you'll need to explicitly push images somewhere accessible.

The Edge Stack service

Edge Stack is deployed as a Kubernetes service. The following configuration will create a service for Edge Stack. It also adds a mapping that will route traffic from

By using Kubernetes annotations, Edge Stack integrates transparently into your existing Kubernetes deployment workflow, so tools such Kubernetes deploy work naturally with Edge Stack.

Save the above YAML into a file called

Starting Edge Stack

We have an Edge Stack service, but we don't actually have Edge Stack running. To do this, we'll need a Kubernetes deployment. If you're using a cluster with RBAC enabled, you'll need to use:

Without RBAC, you can use:

Once that's done, you should see three pods for Ambassador:

$ kubectl get pods
                NAME                                READY     STATUS    RESTARTS   AGE
                ambassador-1378270275-51qns         2/2       Running   0          2m
                ambassador-1378270275-kr361         2/2       Running   0          2m
                ambassador-1378270275-v10x8         2/2       Running   0          2m
                
                $ kubectl get services
                NAME               CLUSTER-IP        EXTERNAL-IP        PORT(S)    AGE
                ambassador         109.168.235.177   a1128c0831f9e...   80/TCP     3m
                kubernetes         109.164.0.1       <none>             443/TCP    9m
                

Edge Stack relies on Kubernetes for replication, availability, and scaling -- making it itself very simple. It also relies on Kubernetes for persistence, so Edge Stack has no database.

Setting

AMBASSADORURL

In order to get access to your microservices through Edge Stack, you'll need an external URL to Edge Stack's service interface. We'll use

We'll need to start by getting the external IP address. You can get the IP address by running

In any case, do not include a trailing 

Testing Edge Stack

We can now speak to the

This will send a request to Ambassador, which then routes the request to the

Diagnostics

Under the hood, Edge Stack relies on Envoy (and its powerful feature set) for routing, TLS, and the like. Includes diagnostics that gives more insight into the Envoy configuration that Edge Stack is managing. These diagnostics are not publicly exposed by default.

You can access the diagnostics by getting the list of Edge Stack pods:

$ kubectl get pods
                NAME                                READY     STATUS    RESTARTS   AGE
                ambassador-1378270275-51qns         2/2       Running   0          2m
                ambassador-1378270275-kr361         2/2       Running   0          2m
                ambassador-1378270275-v10x8         2/2       Running   0          2m

forwarding to port 8877 on one of the pods:

And then visiting http://localhost:8877 in your web browser.

Up Next

In this article, we've shown how you can deploy Envoy as an API Gateway using Ambassador. Envoy has many powerful features such as sophisticated load balancing algorithm, advanced statistics monitoring, and more. Ambassador exposes many of these features through annotations to support use cases. To learn more, read about Canary deployments or see how to use Prometheus to monitor Envoy and Edge Stack.