google_cloud

Yeah AWS Lambda / API gateway and such, I do think in terms of a no manage DB firebase looks better.

The concern I have with all of the serverless approaches comes down to scaling with a team:

* The deployment story is still a bit messy
   * dev environment generally connects to shared resource
   * testing is tricky / non existant
   * debugging tools aren't up to speed.
* doesn't do well with dev/staging/prod completely forgetting as the team scales branch based dev
* less so on AWS to standard DBs, but firebase doesn't have a great story for data analysis yet. 

All that being said, I think serverless looks awesome.
It is quite possibly the future for a large portion of projects depending on size and complexity.
I really want to play with it a bit more and for personal projects will definitely be exploring. 
At this point I would still be a bit to concerned to try to build a buisness around it.

Anyways, just my 2 cents.

Definitely agree with the large team pitfalls of serverless. For small teams though, the serverless can actually help. How many times have you helped front end engineers setup mysql and run db setup & seeds to get a basic version of the code they actually care about up and running? Serverless lets them just git pull, start gulp or whatever frontend server/build tool they're using, and go.

My prediction is that all of those points will be addressed in the next 2 years by Google & AWS and that serverless will be how all new projects are started.

Lastly, it wouldn't be too much of a stretch to start on serverless and transition to a hosted RoR app as the API if your project grew into a large business and AWS/Google didn't have support for the current pitfalls yet.

yep definitely have pain having front end (or mobile Android/IOS devs) setup local dev servers... Can end up wasting a lot of time. Docker can help ease that if you do full dev environment in Docker, but that has some disadvantages.

I do think in the next while we will see most of the pain points resolved for serverless, not entirely sure on the 2 year timeline, but it will evolve quickly.

On the transition totally true, I helped migrate a mobile app with Parse backend to a pure Ruby/Rails backend as Parse was failing to scale to their transaction rate. It takes some work but isn't entirely difficult.. a few months later Parse announced it was shutting down so everyone was happy to be off of it.

Another advantage is Firebase actually handles a lot of offline sync DB issues for mobile devs which is a win as well for early development iterations.

Trek10 who I work with on AWS stuff has been digging more into the testing and multi-environment issues around serverless and hopefully will share some of their solutions shortly ;)