Sun 12 July 2020
The CSOD "Spring" Hackathon was held last week. My team spent a whole day to help me finish an ambitious idea. Though we didn't win the competition in the end, we had a lot of fun and did learn many things.
The AWS lambda serverless with dotnetcore, one of "new" technologies that we used during the event, surprised me with its simplicity and dev-friendliness. It's something could be used for a very quick prototyping in the future.
Here's how we built up a production ready REST backend in a few easy steps.
Before start, we decided our backend stack as asp.net core + MySql. The asp.net core will be transformed to api gateway + lambda, while the MySql will be replaced by the Aurora Serverless. A pure serverless architecture.
First of all, we need to get a local MySQL up and running. It can be easily accomplished by a
docker-compose, nice and clean:
version: '3.3' services: db: image: mysql command: --default-authentication-plugin=mysql_native_password restart: always ports: - '3306:3306' environment: MYSQL_DATABASE: 'db' MYSQL_USER: 'user' MYSQL_PASSWORD: 'password' MYSQL_ROOT_PASSWORD: 'password' volumes: - my-db:/var/lib/mysql adminer: image: adminer restart: always ports: - 8080:8080 volumes: my-db:
Second step, scaffolding the asp.net serverless project. Amazon engineering team provided a set of tools integrated with dotnet-cli, which contains a few of serverless templates.
1) Install global tools set
dotnet tool install -g Amazon.Lambda.Tools
2) Install project templates
dotnet new -i "Amazon.Lambda.Templates::*"
To view the installed lambda templates, run
dotnet new lambda --list
3) Create new project with serverless template
dotnet new serverless.AspNetCoreWebAPI -n serverless
It will create a folder of
example with the structure
example ├── src │ └── serverless │ ├── Controllers │ │ ├── S3ProxyController.cs │ │ └── ValuesController.cs │ ├── LambdaEntryPoint.cs │ ├── LocalEntryPoint.cs │ ├── Readme.md │ ├── Startup.cs │ ├── appsettings.Development.json │ ├── appsettings.json │ ├── aws-lambda-tools-defaults.json │ ├── serverless.csproj │ └── serverless.template └── test/* # test projects folder
The files are almost minimal to develop and deploy the application. Let's quickly walk though the key files
aws-lambda-tools-defaults.json: Used for
dotnet lambdacli command. It defines the necessary deployment parameters.
serverless.template: CloudFormation template that defines the resources to be created in the deployment. By default, it already includes a lambda definition with
proxy+API Gateway trigger.
LambdaEntryPoint.cs: The class wrapped with built-in
APIGatewayProxyFunctionto enable lambda hosting the web app.
LocalEntryPoint.cs: Local asp.net hosting entrance that uses Kestrel.
Startup.cs: Where all the configurations and warmup happen for both local and lambda.
S3ProxyController.cs: A sample controller with AWS s3 client, where we can start to build our app straightforward.
4) Integrate with EF core
As part of dotnetcore family, Entity Framework Core could be easily integrated within an asp.net project.
There is also a dotnet-cli tools set for EF core.
dotnet tool install --global dotnet-ef
The cli sets provides a few useful commands to manage DbContext and DbMigrations. For more information, refer to here.
Once it's connected, put the
ConnectionString into an config file or environment variable. Then goto AWS RDS console, create an Aurora Serverless database and get the connection link. Put the link into your production configuration.
5) Development and deployment
To start coding, we just need to code as our normal asp.net project. Open the project with your favorite IDE.
To deploy your project, just get your AWS api key setup locally and run command:
dotnet lambda deploy-serverless
It will automatically load settings in
aws-lambda-tools-defaults.json (if not, then prompt to ask for input).
At the end of deployment, you'll get an API url:
Output Name Value ------------------------------ -------------------------------------------------- ApiURL https://xxxxxxxx.execute-api.us-west-2.amazonaws.com/Prod/
Till now, our backend project is ready for client consuming.