Dear students,
Welcome to the third exercise for Cloud Computing. Now that you are experts on Docker Compose, we are gonna explore building microservices (kind of, the idea still holds).
During this assignment, you will explore the requirements of minimalizing your containers for deploying microservices! You will also explore using a Load Balancer to orchestrate traffic accross different endpoints.
As previous assignments, we will build on top of your existing code, which you will partition into smaller to generate five different containers. So, what is the task?
To succeed in this assignment, you must do the following:
- Use Docker Compose to orchestrate your five containers and deploy NGINX
to control the traffic between your services based on the request method. The
containers are as follows:
- A container to handle each operation for
/api/books: GET, POST, PUT, and DELETE - A container to handle requests to
/*, i.e., the rendering of the webpage. - A container for NGINX.
- A container to handle each operation for
- Use a Multi-stage Dockerfile to minimize the size of the image.
- Publish to a public container registry (e.g., Docker Hub) your images.
- You must create all images for
x86_64(i.e.,amd64). - You must use a DB: you can run it as a container or install it directly in your VM.
- Create unique code-bases to handle each operation (GET, POST, PUT, DELETE, and server-side rendering) by decoupling the current code structure.
- The URI for MongoDB must be given via an environment variable.
- The traffic must be redirected via NGINX by checking the request's method.
- Do not modify the headers inserted by NGINX.
- The automated tests from Exercise 1 must pass.
- Using an insolated MongoDB instance, the submission server will download, configure, and run all of your images. (10 pts).
- Using the isolated instance, the submission server will perform tests to the different available endpoints (45 pts).
- Using your VM, the submission server will perform tests with the given endpoint and check that you are running NGINX with proper configuration (45 pts). Please make sure the port in the given port matches the exposed one for NGINX.
The starting point is the code from your last exercise and your existing Docker Compose file. From there, you will have to expand such file to include:
- The configuration for NGINX. NGINX can be configured using an
nginx.conffile that you must pass to the respective container by mounting a volume into the respective path. Please look at this tutorial by Digitial Ocean, which explains the structure of the file. Since we want to redirect based on request's methods, this page could provide insightful information on how to configure your$location. Hint - Configure your other services to start in a timely manner so they are reachable once NGINX is up.
- Use MongoDB to run before everything starts.
Since we are looking to optimize the size of a microservice's container, it might be useful to explore multi-stage builds to reduce the size of the final container. You can read more here on how to create a multi-stage build.
Basically, we want you to first build your executable: download necessary packages,
compile your code, and deploy on a thinner container. Given the multi-stage containers,
you can copy the compiled code (and necessary dependencies) from a previous stage
to the current one (a smaller one). Usually, people use an alpine version of linux
to deploy lightweight containers. That means, you build your code using the current heavy-
weight Golang container, you copy the final binary, and the required files (for
rendering HTML) into the smaller stage, which generates a tiny image perfect
for microservices.
If you want to improve your code even more and remove file dependencies, Golang
provides a neat feature called go-embed
to compile a given folder into the final binary to remove dependencies with the
host filesystem. Since I cannot check whether you are using such feature, I leave
it to you to explore!