This is meant to be the most efficient, scalable, and fast production-ready serverless REST API backend which provides CRUD operations for a MongoDB collection!!
It is meant to be optimized for:
Written in Rust with most optimized compiler settings and zero "virtual machine bloat" that exists with languages such as python, node, or go.
This project packages the Rust server as AWS Lambda functions where you pay for only the milliseconds or server time when your code is executing. As opposed to other architectures that involve running a server 24/7, serverless functions do not incur a cost when the system is not in use, and there is no more headache of configuring load balancing correctly.
-
Rather than a "get all" method that, this project provides a "get single" and "get paginated list" which uses cursor pagination
-
Includes a "move data to lake" to lake function which offloads old documents from the mongo "application database" over to a mongo "data lake" for long-term storage and analytics purposes.
This is a Rust project. Please first install cargo and serverless.
Running Tests:
cargo test
Running Locally (Note - not working for me...)
sls invoke local -f function_name
Deploying:
sls deploy
Note: You will need to first configure your aws cli before deploying:
aws configure
Modify constants defined in .env to your liking.
You may also want to update your resource name.
This must be done manually at the current time...
Do a project find-and-replace for these things:
- cat -> my_new_resource_name
- Cat -> MyNewResourceName
- CAT -> MY_NEW_RESOURCE_NAME
You can also add / remove fields on the Cat struct defined in: ./src/cat/cat.model.rs
We usually recommend using the plural form of the resource for which you are building this CRUD API.
For example, if each document in your collection represents a "cat", then the url endpoint should be "cats".
POST to: https://your_url/cats with cat details
GET: https://your_url/cats/123123123 where "123123123" is the document _id for the cat object to get.
GET: https://your_url/cats/123123123 where "123123123" is the document _id for the cursor document. Returns up to 10 cats after the cursor doc.
By default these get the full cat document. You can optionally pass a body with these requests which is used in the mongo "find" call the projection object, allowing you to call for a subset of the cat document (by specifying either the fields you want included or excluded in the data to get).
POST: https://your_url/cats/123123123 where "123123123" is the document _id for the document to update. Takes an object of fields to update and leaves existing fields on the document.
DELETE: https://your_url/cats/123123123 where "123123123" is the document _id for the document to delete.
POST: https://your_url/cats/move-stale Copies docs older than X days (default 30) to mongo data lake and deletes them from the db collection.
This project has 100% code coverage and great assertions!
It uses a mock mongo server so that you can run the tests without calling out to an external database and without having to directly mock the functions in the external mongo crate.
This project also has some end-to-end tests that ping each endpoint and fail if responses are not fast enough.
Uses the [artillery] project and defines e2e tests in yaml.
// TODO - get some real data for these functions!!
Feel free to open issues / PRs if you have ideas on how to improve this project