Needs advice

I'm currently developing an app that ranks trending stuff ( such as games, memes or movies, etc. ) or events in a particular country or region. Here are the specs: My app does not require registration and requires cookies and localStorage to track users. Users can add new entries to each trending category provided that their country of origin is recorded in cookies. If each category contains more than 100 items then the oldest items get deleted. The question is: what kind of database should I use for managing this app? Thanks in advance

4 upvotes·157.1K views
Replies (1)

I think your best and cheapest choice is going to be MongoDB, Although Postgres is probably going to be the more scaleable approach, you likely have a good idea of how you want to present your data, and the app seems small enough that you shouldn't need to worry about scaling issues. It also sounds like your app can grow in a linear capacity based on the number of users, and the amount of data, which is the perfect use-case for noSQL databases (linear, predictable scaling).

Correct me if I have any of these assumptions wrong. 1. You're looking to have a relatively high-read with a lower write volume 2. Your app is essentially a list of objects that can belong to a category 3. users can create objects in this list.

I think Mongo is going to be what you're looking for on the following basis: 1. you absolutely need a database that is shared by all users of your app, therefor IndexedDB is out of the question. 2. You have semi-structured data 3. you probably want the cheapest solution.

I think Postgres is wrong for the following reasons: 1. your app is pretty simple in concept, SQL databases will add unnecessary complexity to your system, either through ORMs or SQL queries. (use an ORM if you go with SQL) 2. Hosting SQL databases for production is not cheap! the cheapest solution I know of for Postgres is ElephantSQL. It provides 20MB for free with 5 concurrent connections, you should be okay to manage these limitations if you decide to go Postgres in the end. Whereas mongoDB Atlas has some great free-tier options.

Although your data might be easier to model in Postgres, you can certainly model your data as a single list of items that have a category attached.

I don't want to officially recommend another tool, but you should really checkout prisma, firebase, amplify, or Azure App Services for this app! Just go completely backend-less [Firebase] [Amplify] [Prisma] [Azure App Services]

6 upvotes·2 comments·151.1K views
May 18th 2020 at 9:58AM

Thanks for the answer

François Delpierre
François Delpierre
May 27th 2020 at 1:15PM


Sorry Edwin, but I can't agree 100% for the following 2 reasons :

- MongoDB is far more scalable than PostgreSQL because it supports sharding. Running a cluster of 12 mongodb nodes containing 4 shards as 4 replica sets of 3 replicas is relatively easy with Mongo. But in your case, it's not so relevant as it's always better first to upgrade hardware before thinking about enabling sharding.

- PostgreSQL DO supports JSON / document based (NoSQL) storage for ages (it was probably the first RDMS on the market to support documents storage), and is even much faster than MongoDB in many cases. So, the benefits is that you can have in the same DB relational and document based data.

For all the rest, I fully support your answer, including the preference for MongoDB. While it might be a new language to learn, this map-reduce way of building queries in MongoDB is very powerful, and can provide very fast and lightweight answers that are so crucial for today's online apps.

Avatar of Edwin Jackson