apls | Prolog Logic Server , core components | Math library

Firebase is an awesome BaaS - Backend as a Service. If you have no need to create a custom backend then you have no need. This just means your web app may not need the added complexity. Here are reasons someone would create their own:

• Requests get Cached - If someone using your web app makes an API request directly to your Firebase API, some other user may make the same request seconds later. If you had a Node.JS server making that request to FireBase those 2 requests would only be 1, as your server would've cached the response.

• Cost - using the example I gave above imagine that number being multiplied by 20 you can begin to see how your costs would differ, as when it comes to a BaaS you're paying for bandwidth. Now say you want your web app to be used on different continents, does Firebase allow GCP Regions (this allows you to have a server instance in places like US-WEST, US-EAST, EU-London)?

• Complex Data Models - One of the problems with some BaaS' is the issue of related data. While you may have simple relationships in your current app, imagine a web app that has much more complexity around related models. Say this is the data you're needing for an angular view:

  1. User
  2. Posts (last 10)
  3. Likes (last 10
  4. Favorites (last 10)

  In a lot of cases it would be more beneficial for your server to do a much more complex query, gather this data and send it back in 1 requests versus coding 4 different requests in the angular client. Imagine if this continues how large the angular code base can become, this ties back to the point about costs. Now imagine paging through just the Posts, you'd have to get pretty creative with your Firebase requests.

These are the things to consider. If they don't apply, then they don't apply, and if they do YOU'LL know.