Which nosql database to choose for the project's highload?

Question

NoSQL

I haven't worked with nosql, always with mysql/postgreSQL so don't hit it
Need to develop crm.
It collects a lot of information, but only conditionally four types of information (tables)
From them, you need to generate reports with a large number of associations of these tables.

Answer 1

1. If you need joins, you don't need nosql.
2. Imho it is necessary to develop any system with 3nf in classical relational basis. When both the architecture and the real load distribution are clear, you can already deal with denormalization.
3. Don't trust anyone. Don't believe me either. Do personally load tests of a specific version of a specific storage close to your task. The results may (un)pleasantly surprise you.
My monga died very quickly. I did not see people with a live loaded project and mongo.
You might be lucky ;-)

Answer 2

I think mongo is quite up to the task. It is fast and handles large amounts of data.

Answer 3

postgres
Until you understand what the structure of the project is, you can store data in json
When everything settles down, you normalize

Answer 4

I also advise you not to ignore Elasticsearch . Very interesting selections can be made. At first, you will have to blow up the brain a little, but then it will go like clockwork.
It took me about two months of leisurely study to comprehend everything. Recoil speeds are fantastic. There was just a choice between mongo and elastic. As a result, elatsik won with novelty. Well, at one time I used lyusina very actively (a search engine that uses elastic), about 10 years ago, so I had a rough idea of ​​what it was.
The only thing I recommend is to immediately take a client library with query automation, bare queries are still a pleasure. I am using elasticsearch-dsl for python.