Below is how to run my solution, along with the answers to various questions.

From the root dir....

To build the docker app image:

./build

To start the containers:

./start

To migrate the database

./migrate

You might want to create a virtual env. Make sure that the requirements in the requirements.txt and installed e.g.

pip install -r app/requirements.txt

To start sending data

./send_data

Visit http://localhost:8088/?pgsql=db&username=xbird&db=xbird&ns=public

I chose a relational sql database. While nosql databases can be useful in certain situations, relational databases have been depended on and optimized for decades. The ability to define a schema, and change that schema over time, means you can reliably maintain data consistency. The data that is defined in the protobuf format is easy to map into a relational schema.

I chose postgres because it is opensource and has a huge active community. It has great features such as json columns which allow for schemaless data if needed. (I am also familiar with it!)

I would batch the samples and process them asynchronously using jobs and workers (e.g. celery)

I would use logging aggregator such as prometheus or papertrail. I would also have alerts set up do monitor things such as db cpu usage, etc.

As I have not used python in a long time, I found this more time consuming than it should have been. It took me a while to set up database migrations, etc. Most of the tools I had to use were not familiar to me. Thus I did not have the time to do many things I would have liked too e.g.

• Add various database indexes.
• Handle errors.
• Tests! (I usually use extensive testing, but haven't used a python test framework in over 2 years.)
• Kubernetes (Docker compose is a simple easy alternative - but not scalable.)
• Location type samples.
• etc.