For the backend we chose Express, a Node.js based web application framework, mainly because of its simplicity, and we wanted something that is less complex for this particular project.

For the bidirectional event-based real-time communication we would rely on web sockets, and we chose Socket IO library, due to its ability to fail over to other protocols when web sockets are not supported, automatic reconnection, proxy and load balancer support and most importantly the broadcast support.

This avoided so much boilerplate code, and it also integrated seamlessly with the Express server. MongoDB is used to store user and configuration data, as well as queue logs. For actual management of the queue, we used a combination of Redis Hash and Sorted Set with Lua scripts to run many queries in transaction.

For the front-end, we went with Flutter for its cross-platform support, and it also integrated well without socket IO backend. We found it perfect to use the Picture-in-Picture feature to show the queue status if the user wants to use another application.

This would enhance the user experience because the user does not need to be locked on to a single screen to know the current queue status.

We also added various configuration parameters to control the queue which could prioritize the user’s position in the queue based on their token number, whether they missed their turn or not, whether they are a late-joiner, etc.

The front-office admin is provided with two controls, one to go to the next user and one to skip the current user, which will result in moving the user down based on the queue configuration. This service, paired with booking and video call microservices, pretty much covered the core features of the project we were developing..