On an architectural level The solution makes use of event sourcing to keep track of the heartbeat events and CQRS for separating write and read operations. This enables the system have an accurate log of the heartbeat -ping messages- from the vehicles. More over it would introduce sympathy between the system and the domain which is inherently event-based.
- Flask as a micro framework for building micro-services.
- SqlAlchemy as an ORM, with Flask-Migrate for database migrations.
- unittest python module for writing unit tests and integration tests.
- nose test module as a test runner.
- postgres as a database.
And the whole back end is orchestrated by docker swarm
NOTE: Ideally I would have used different databases for storing events (preferably elasticsearch) and
a relational database for the association between customers and vehicles.
- Angular as a framework to build the single page app
- Jasmine and Karma were used to write unit tests and integration tests.
- unittest python module for writing the test cases.
- Selenium for writing the test cases logic.
For continuous integration / continuous delivery Travis were used for running unit tests and building and pushing docker images to docker registry (dockerhub)
In the case of the back end the a cloud provider would be providing the infrastructure the app would be running on including the virtual machines across multiple regions which would be used as nodes in our clusters, network interfaces for communication between nodes and in some cases we would be hosting SSL certificates and managing our DNS.
On the front end side we can use the cloud to provide us with CDN hosting service to introduce high availability and faster respond time to users.
Please find instructions on deployment steps below:
In the case of using the serverless architecture for for the same challenge the cloud provider would take care of orchestrating the services (functions) either by templates or a web console. auto scaling would be handled by the cloud provider as well. To expose the serverless functions to to be invoked we can use an api gateway.
For data storage we can use elasticsearch for storing events and a relational database (possibly postgres) for storing relational data like customer and vehicles data. We have the option to use self-managed database or use database as a service.
We would have 2 serverless functions one for writing events (the heartbeat function) and the other for searching/filtering (the search function)
