10 March, 2020

Azure Web App for Containers + Docker Compose

There is a lot of talk around containerization nowadays. Everyone seems to be jumping into this bandwagon and thinking in terms of dockers and containers only. Cloud has also given a big boost towards it and made it even more popular by providing the platform with lot of flavours for containerization. Azure "Web App for Containers" in one such service which though not very popular among developers like Azure Kubernetes Service and Service Fabric but has a lot of sweet potentials.

We will investigate a great feature of this service today. We will create a sample Web app and an API app, containerize those and will leave it to the "Web app for containers" service to do all the necessary plumbing work and making sure our application is up and running. To make it more fun, we would only make use of Visual Studio  mostly in our demo and will only switch to Azure for customization purpose. Sounds interesting... Let's get started.

Prerequisites:
1.    Azure Subscription (Free subscription of 30 days will also do)
2.    Microsoft Visual Studio 2017
3.    Docker Desktop for Windows - this should be installed first, up and running
4.    Basic C# programming knowledge
  • Open VS 2017 and select File >> New >> Project. Select ".Net Core" and "Asp.Net Core Web Application" as the template. I have given the name "MyWebApp" as the project name, but you are free to give anything of your choice.
  • Select the following in the next screen. Since this is going to be web application, hence the web application (model-view-container) option is selected. I have taken Asp.Net Core 2.1 as the latest version. Leave other parameters as default and click OK to create the project.
  • Right click on the project to add container orchestrator support as below.
  • Select "Docker Compose" from the drop-down and click OK
  • Select target OS as "Linux" and click OK to complete the process
  • Now let's see how VS has configured our web project. A Docker file got added along with a new project name docker-compose. The magic here is that our application now has got docker along with orchestrator support and we do not need to change a single line of code in any of the files. VS has done all the necessary handshaking.  

Let's add the API project. 
  • Right click on the solution and add a new project. I have given the name "MyApiApp" but you are free to give any name of your choice
  • The time select the API as a template and un-check the HTTPS configuration checkbox as we would not require it this time.
  • Now follow the same steps you have done to add docker container orchestrator support in this project as well. Remember... it's the exact same steps. If you have performed it right, your project structure should look similar to this

  • And the docker-compose.yml should look like this.

Now let's introduce some activity in the boilerplate projects.
Open the Controller/ValuesController.cs file under the API project and replace the first Get() function with the below code block. This is just returning the current short time to the calling application.


// GET api/values
[HttpGet]
public ActionResult<string> Get()
{
   return "" + DateTime.Now.ToShortTimeString();
}


Let's call this api from our web application and display the value accordingly.
Open the Controller/HomeController.cs file under the Web project and replace the Index method with the below code block. Here we are just calling the api app get() method to get the current time and storing it into a ViewData.

public async Task<IActionResult> Index()
{
   using (var client = new System.Net.Http.HttpClient())
   {
      var request = new System.Net.Http.HttpRequestMessage();
      request.RequestUri = new Uri("http://myapiapp/api/values/"); //Docker compose will automatically map the underlying IP
      var response = await client.SendAsync(request);
      ViewData["msg"] = "The current time is : " + response.Content.ReadAsStringAsync().Result;
   }
   return View();
}

Open the Views/Home/Index.cshtml page and display the ViewData message just after the Carousel
as below. But yes again you are free to add this message anywhere in the page.


Done with the local setup. Now press F5 to build and run the solution. This will now build container images for both of your web and api application and would launch the project as a container application in a browser. The time (as highlighted below) is coming from the API get() call. Refresh the page and you would see new time every minute.


Great... our application is now containerized and working seamlessly together. What we have done? simply nothing... in fact no docker knowledge needed. All the heavy-duty work is happening behind the scene 😊 

Let's host it in the cloud. Here we need to perform two things
1.    Host the docker images in Azure Container Registry (ACR)
2.    Create an Azure Web App for Containers service to run those images
Make sure you are connected to your Azure subscription from within you Visual Studio IDE.
Right click on the API project and hit "Publish" from the context menu and follow the below steps
  • Select "Container Registry" and then "Create New Azure Container Registry". Click Publish.
  • All the fields will be automatically populated. I preferred to leave the DNS prefix as is (you can change it as per your convention) for this demo and make sure the right subscription is selected (in case you have multiple subscriptions). You can select an existing resource group or create new. I have created a new one. Select he SKU as Basic as this is enough for this demo and select the location of your choice. Click "Create"
  • Now have some patience. Azure will create the ACR for you and would also publish the docker image related to the API app into it. This will take few minutes.
  • Once done, do the same operation for the Web application as well. But this time select the existing container registry you created in the step one.
  • Now Login to Azure (atlast 😉)
  • Find your resource group (in my case it's my-web-app-rg) and click to see the details in it.
  • You will find the container registry you created
  • Click the container registry. Then click the Repositories (under services) option from the panel. You must see the two docker images you had just published from your visual studio


Now we are all set to create a web app which will run these images. No code or DLLs to deploy. The docker images are self-contained and the rest will be taken care by the azure platform. Sounds great. So, let's create one.
  • Search for "web app for container" in the Azure search bar and select it

  • Fill up the form with the necessary details and click "Next:Docker".
  • Select the values as shown below. The "docker-compose.yml" configuration file you will find in your project directory. But before uploading the file you need to make a very important change. Replace the ${DOCKER_REGISTRY-} variable with the Azure container registry login server name which you will find under the overview tab of the ACR (in my case it's myapiapp20200120095948.azurecr.io).
         Hence the updated docker-compose.yml stands like below

 
 Click "Review + create". Then click "Create". This will fire up the "Web app for containers" setup. Wait for few minutes.
  • Go to your resource group to find the below services in there
  • Click the web app service and hit the Url from the Overview page which will launch your web application in the browser with the time stamp picked up from your API app


Voila!... your web application is now running two container images behind the scene. You can now leverage the power of both docker and azure web app in building a robust and scalable application which is totally based on PaaS. Isn't it a great service?

You can also fully automate the above processes with DevOps. 👍

Congratulations! for coming this far. Hope this article will boost up your self-confidence and would encourage you to further explore this service.

Do share with me about your experience and what you have built upon this foundation. You can take it up to any level and integrate. I would love to hear from you.


No comments:

Post a Comment