To build executable jar file, use:

./gradlew clean bootJar

To run test:

./gradlew clean test

To do both:

./gradlew clean build

To run the service:

./gradlew clean bootRun

or if you use IntelliJ, it will detect application entry point. But make sure you have a MongoDB container is running on your local environment. You can do that by:

./gradlew composeUp

The most significant property of the requirement is high write workload. I concerned about several Relational DBMS I've worked with such as MySQL, PostgreSQL but seems they're not really fit for a high write workload that have to finish in 9 hours.

Almost SQL DBMS will validate data before insert to storage space. Default isolation level of MySQL is snapshot, PostgreSQL is read committed. Of course, we can do turning for it but why don't we choose a NoSQL DBMS that has good support for high write.

So I chose MongoDB, although I haven't worked with it. But after Googling, I found that MongoDB support high write through:

• Read uncommitted as default isolation level https://docs.mongodb.com/manual/core/read-isolation-consistency-recency/#read-uncommitted
• MongoDB supports flexible schema, so it doesn't have many constrains on field validation, so is doesn't slow down insert operation.

About distributed database, it's risk to do in 9 hours.

Application design is simple. I have a Java web application that connect to a MongoDB to store data. The web application can scale by add more instances. The MongoDB can scale by add more resource, use sharding or replication.

In MongoDB database, I created a collections that has schema following class AirDeviceData. I also added index for these field: deviceId, timestamp. Because these field will be used to query and sort data. For more detail, please check docker/init-mongo.js.

I divided code to several packages base on its responsibility.

• When I meet a class that seems complicated, I will try to refactor it by split out functionality.
• Small class made writing unit test easier.
• No complicated static method.
• Refactor code base on unit test.

• Preparation: 4 hours (research)
• Coding: 3 hours (including tests)
• Documentation: 2 hours
• Verify: 1 hours
• Grand total: 10 hours