aisus / interview-timeseries

interview-timeseries

An example application built with .NET and TimescaleDB.

This application processes and stores metric events emitted via websocket and exposes a simple JSON API to explore the data.

Architecture

This app consists of two services: api and listener.

api is an ASP.NET Web API intended to view the stored metrics.

listener is a worker service which listens to websocket events in background and writes data to a TimescaleDB.

How to run

Using Docker

• Clone this repository
• Run the project using docker-compose:

docker-compose -f .\docker-compose.yml up --build -d

• docker-compose will create three containers: api, listener and timescaledb. Database uses the default credentials (timescaledb password).

Access the SwaggerUI via following URLs:

• Swagger - http://localhost:8000/swagger/index.html

Running in VSCode

• Clone this repository

• Start the TimescaleDB separately using docker-compose:

docker-compose -f .\docker-compose.yml up --build -d timescaledb

• If you wish to use a different database, update connection strings for TimescaleDB to point to your local database server in api appsettings.Development.json and listener appsettings.Development.json

"ConnectionStrings": {
    "TimescaleDB": "host=localhost;port=9000;User ID=timescaledb;Password=password;Database=timeseriesstorage;"
  }

• Database migrations would run on the first startup.

• Run either api, listener or both using configurations in launch.json

Access the app via following URL:

• Swagger - https://localhost:7055/swagger/index.html

Running in other IDEs

• Do the same steps as explained for VSCode
• Configure your projects to run using Development environment

Future proposals

To scale this application for production, I propose following:

• Split listener into two separate services: listener and worker
• listener would still handle the websocket connection, while worker would be responsible for saving the data to TimescaleDB.
• Connect the services via a message queue, e.g. RabbitMQ
• Batch / bufferize incoming requests in the listener side.