Skip to content

Latest commit

 

History

History
163 lines (116 loc) · 9.13 KB

README.md

File metadata and controls

163 lines (116 loc) · 9.13 KB
Raw

Kubernetes Distributed Calculator

This sample shows method invocation and state persistent capabilities of Dapr through a distributed calculator where each operation is powered by a different service written in a different language/framework:

  • Addition: Go mux application
  • Multiplication: Python flask application
  • Division: Node Express application
  • Subtraction: .NET Core application

The front-end application consists of a server and a client written in React. Kudos to ahfarmer whose React calculator sample was used for the client. The following architecture diagram illustrates the components that make up this sample:

Architecture Diagram

Prerequisites

In order to run this sample, you'll need to have an Dapr-enabled Kubernetes cluster. Follow these instructions to set this up.

Running the Sample

  1. Clone the sample repository 
    git clone https://github.com/dapr/samples.git
  2. Navigate to the deploy directory in this sample directory: cd deploy
  3. Follow these instructions to create and configure a Redis store
  4. Deploy all of your resources: kubectl apply -f ..

    Note: Services could also be deployed one-by-one by specifying the .yaml file: kubectl apply -f go-adder.yaml.

Each of the services will spin up a pod with two containers: one for your service and one for the Dapr sidecar. It will also configure a service for each sidecar and an external IP for our front-end, which allows us to connect to it externally.

  1. Wait until your pods are in a running state: kubectl get pods -w
NAME                                    READY     STATUS    RESTARTS   AGE
dapr-operator-775c97497c-p92mf          1/1       Running   0          134m
dapr-placement-58c7d5f9cf-l9wcv         1/1       Running   0          134m
dapr-sidecar-injector-5879986bdc-nwdps  1/1       Running   0          134m
calculator-front-end-7c549cc84d-m24cb   2/2       Running   0          3m
divideapp-6d85b88cb4-vh7nz              2/2       Running   0          1m
multiplyapp-746588586f-kxpx4            2/2       Running   0          1m
subtractapp-7bbdfd5649-r4pxk            2/2       Running   0          2m

  1. Next, let's take a look at our services and wait until we have an external IP configured for our front-end: kubectl get svc -w

    NAME                          TYPE           CLUSTER-IP     EXTERNAL-IP     PORT(S)            AGE
dapr-api                      ClusterIP      10.103.71.22   <none>          80/TCP             135m
dapr-placement                ClusterIP      10.103.53.127  <none>          80/TCP             135m
dapr-sidecar-injector         ClusterIP      10.104.220.35  <none>          443/TCP            135m
addapp-dapr                   ClusterIP      10.0.1.170     <none>          80/TCP,50001/TCP   2m
calculator-front-end          LoadBalancer   10.0.155.131   40.80.152.125   80:32633/TCP       3m
calculator-front-end-dapr     ClusterIP      10.0.230.219   <none>          80/TCP,50001/TCP   3m
divideapp-dapr                ClusterIP      10.0.240.3     <none>          80/TCP,50001/TCP   1m
kubernetes                    ClusterIP      10.0.0.1       <none>          443/TCP            33d
multiplyapp-dapr              ClusterIP      10.0.217.211   <none>          80/TCP,50001/TCP   1m
subtractapp-dapr              ClusterIP      10.0.146.253   <none>          80/TCP,50001/TCP   2m

    Each service ending in "-dapr" represents your services respective sidecars, while the calculator-front-end service represents the external load balancer for the React calculator front-end.

    Note: Minikube users cannot see the external IP. Instead, you can use minikube service [service_name] to access loadbalancer without external IP.

  2. Take the external IP address for calculator-front-end and drop it in your browser and voilà! You have a working distributed calculator!

    For Minikube users, execute the below command to open calculator on your browser

    $ minikube service calculator-front-end

Calculator Screenshot

  1. Open your browser's console window (using F12 key) to see the logs produced as we use the calculator. Note that each time we click a button, we see logs that indicate state persistence:
Persisting State:
{total: "21", next: "2", operation: "x"}

total, next, and operation reflect the three pieces of state a calculator needs to operate. Our app persists these to a Redis store (see Simplified State Management section below). By persisting these, we can refresh the page or take down the front-end pod and still jump right back where we were. Let's try it! Enter something into the calculator and refresh the page. The calculator should have retained the state, and your console should read:

Rehydrating State:
{total: "21", next: "2", operation: "x"}

Also note that each time we enter a full equation (e.g. "126 ÷ 3 =") our logs indicate that we're calling our to a service:

Calling divide service

Our client code calls to an Express server, which routes our calls through Dapr to our back-end services. In this case we're calling the divide endpoint on our nodejs application.

Cleanup

Once you're done using the sample, you can spin down your Kubernetes resources by navigating to the ./deploy directory and running:

kubectl delete -f .

This will spin down each resource defined by the .yaml files in the deploy directory, including the state component.

The Role of Dapr

This sample demonstrates how we use Dapr as a programming model for simplifying the development of distributed systems. In this sample, Dapr is enabling polyglot programming, service discovery and simplified state management.

Polyglot Programming

Each service in this sample is written in a different programming language, but they're used together in the same larger application. Dapr itself is language agnostic - none of our services have to include any dependency in order to work with Dapr. This empowers developers to build each service however they want, using the best language for the job or for a particular dev team.

Service Invocation

When our front-end server calls the respective operation services (see server.js code below), it doesn't need to know what IP address they live at or how they were built. Instead it calls their local dapr side-car by name, which knows how to invoke the method on the service, taking advantage of the platform’s service discovery mechanism, in this case Kubernetes DNS resolution.

The code below shows calls to the “add” and “subtract” services via the Dapr URLs:

const daprUrl = `http://localhost:${daprPort}/v1.0/invoke`;

app.post('/calculate/add', async (req, res) => {
  const addUrl = `${daprUrl}/addapp/method/add`;
  req.pipe(request(addUrl)).pipe(res);
});

app.post('/calculate/subtract', async (req, res) => {
  const subtractUrl = `${daprUrl}/subtractapp/method/subtract`;
  req.pipe(request(subtractUrl)).pipe(res);
});
...

Microservice applications are dynamic with scaling, updates and failures causing services to change their network endpoints. Dapr enables you to call service endpoints with a consistent URL syntax, utilizing the hosting platform’s service discovery capabilities to resolve the endpoint location.

Simplified State Management

Dapr sidecars provide state management. In this sample, we persist our calculator's state each time we click a new button. This means we can refresh the page, close the page or even take down our calculator-front-end pod, and still retain the same state when we next open it. Dapr adds a layer of indirection so that our app doesn't need to know where it's persisting state. It doesn't have to keep track of keys, handle retry logic or worry about state provider specific configuration. All it has to do is GET or POST against its Dapr sidecar's state endpoint: http://localhost:3500/v1.0/state/${stateStoreName}.

Take a look at server.js in the react-calculator directory. Note that it exposes two state endpoints for our React client to get and set state: the GET /state endpoint and the POST /persist endpoint. Both forward client calls to the Dapr state endpoint:

const stateUrl = `http://localhost:${daprPort}/v1.0/state/${stateStoreName}`;

Our client persists state by simply POSTing JSON key-value pairs (see react-calculator/client/src/component/App.js):

    const state = [{ 
      key: "calculatorState", 
      value 
    }];
    
    fetch("/persist", {
      method: "POST",
      body: JSON.stringify(state),
      headers: {
        "Content-Type": "application/json"
      }
    });