Hashicorp projects also seem very well done too especially given how extensible they are.

(Btw there are 3.5mm to lightning adapters that let you charge while listening, just not made by Apple)

> I'd suggest porting to key/value pair DBs like mongo or firestore.

Why? The data set feels pretty "relational" to me (contests have divisions, divisions have scorings and results, results have players, players have sponsors) and seems to fit SQL. To be fair though, I've not used Mongo or Firestore or a NoSQL db in general before.

The "scorings" part is the only thing that tripped me up designing a schema since it varies between divisions in a contest as well as between contests themselves, so I ended up w/ a JSONB column for scoring rubric for divisions and then in the results table there's both final_score and scoring_breakdown columns, the beakdown being another JSONB column w/ the individual category results.

I'm putting together a small personal project related to one of my hobbies, basically a tiny web app. It will have a small Postgres DB, a basic Flask API, and then eventually a TypeScript frontend. The MVP is essentially creating a query-able backend for contest results, though eventually I'd like to expand it to a whole management platform for these contests. The events happen infrequently (~3 times a year at most) so the data grows in (small) bursts and is otherwise static. At this point it's entirely text and number data as its currently managed through spreadsheets.

For deployment, I'm considering:

* AWS RDS + EC2 (or the GCP equivalent) for the db + API * Fly.io (or another provider in this space (Heroku?)) because deploying at the edge is appealing and their DB pricing seems reasonable * Fully manual via VPS / dedi hosts like OVH/Vultr/Scaleway/etc

AWS/GCP is appealing because of familiarity through work and it offloads having to manage the databases myself and the data centers are closer to our user base. EC2+RDS works out to ~$20/mo for EC2 (t4g.small) + ~$25 for RDS (db.t4g.micro), GCP is about half that (~$25) for an e2-small and db-f1-micro.

Fly.io is similarly appealing with the benefit of edge deployments, not that my application is particularly latency sensitive. Fly.io would be between ~$7 (free tier compute + basic postgres) and ~$20. Heroku would be between free and $16/mo (for the $7 hobby dyno and $9 hobby basic psql)

OVH et. al are appealing on a cost basis, but feel like they would require a lot more work to ensure similar reliability to the other options. Kimsufi would be between $10 - $40/mo depending on the data center location (Canada or France), Vultr (VPS) would be $10-15/mo, Scaleway Dev (VPS) instances would be $10-$20/mo, Hetzner (VPS) would be $10-$15 (though the downside being their data centers are in Germany and Finland while the majority of our users are in the USA). Specs seem better for the money for these but at the cost of having to manage all of the infrastructure myself.

Comments URL: https://news.ycombinator.com/item?id=28881843

Points: 14

# Comments: 20

[0] http://www.vapoursynth.com/