techx/quill

Docker and/or Firebase?

Opened this issue · 8 comments

aaronosher commented

Hey, so I'm already Dockerizing quill (pretty much done with that) and I'll submit a PR once I'm happy. I was wondering would be interested in a firebase port, using Real-time Database or Cloud Firestore instead of MongoDB. I probably wouldn't port it over to use the firebase SDK on the frontend, just to use cloud functions instead, but that's probably dependent on how much time I have to kill.

Thoughts?

ravirahman commented
aaronosher commented

That's fair, and I've been digging a bit more around the source code and can see how for larger events it could become a fairly large issue. Moreover, in trying to adapt quill to better suite HackNotts, I've run into a handful of quirks that I can't live without fixing. I'm probably going to take a week or so at some point in July/August and port quill over to Angular 6 with Cloud Firestore and Cloud functions. Angular 6 mainly because I never worked with AngularJS and don't really see a point in learning, and partially because the angular SDK for firebase is only for Angular 2+. Firebase because I'm used to it, and it's cheap.

aaronosher commented

I'll PR my Dockerfile/Docker-Compose config in the next week or so.

  • 👍1
ehzhang commented

I've been digging a bit more around the source code and can see how for larger events it could become a fairly large issue.

What particular points do you see being performance bottlenecks? I don't see a reason why anyone would or even should encounter performance problems for a fairly straightforward/lightweight CRUD application with barely any models, even if you have multiple thousands of users and thousands on thousands of hits.

aaronosher commented

What particular points do you see being performance bottlenecks?

The bottlenecks come in the sense that Cloud Functions aren't optimized to be used as a REST API for everything, and it's a lot more costly compared to hitting Firestore directly through the SDK–especially as you get charged twice, once for the function call, and then also for the Firestore or RDB hit. In any app where I use cloud functions as an API point is notably slower than direct Firestore SDK calls.

ehzhang commented

Ah, my misunderstanding - I read your message as if this application as is (no cloud functions) had performance bottlenecks.

jay-khatri commented

@aaronosher any developments on dockerizing quill? If you're able to share a Dockerfile, that'd be great!

aaronosher commented

@jay-khatri My summer seems to have run away from me. If you head over to the hacksocnotts fork on the feature/dockerize branch, there are new Dockerfile, docker-compose.yml and .dockerignore files. As far as I can tell they work as intended. Configure the docker-compose.yml file to your liking and run $ docker-compose up it should work.

I've linked it here: https://github.com/HackSocNotts/quill/tree/feature/dockerize

Edit: I also just submitted a PR to add the Docker functionality #86.