ElasticSearch和Mysql

Question

I have to decide in my next app between NoSQL and Relational Database. The App will be written using ReactJS and ExpressJS.

I have some relational data like videos that have tags and users that like videos and I need to keep track the users that like the videos to disable multiple votes on the same video.

A little image of the beginning of the database :

So It seems that NoSQL doesn't fit well for that. But I want to have the ability to filter the videos and ElasticSearch seems to do the job well with high flexibly but the sync between Mysql and ElasticSearch is harder to maintain I didn't find a lot : This one seems good jprante/elasticsearch-jdbc.

But I keep asking myself some questions :

• How the relation will be imported in elastic search ?
• If I only need to filter some part like videos do I need to sync all the tables ?

So do I need to go with NoSQL or not ? I never use NoSQL or ElasticSearch before.

Answer 1

It seems like you know how to use relational databases, but when you consider NoSQL you come to it from a relational point of view.

The NoSQL world is big and different databases offer different advantages and disadvantages.

For example: Using solr/elasticsearch/mongo could allow you to store data and query it rather efficiently, while other databases, such as key-value big-data databases (cassandra and Hbase, to name a few) will provide even better performance, but you will be limited to "exact-match" queries.

Personally, I'm a strong believer in using relational databases as long as the amount of data being stored isn't too much for it to handle. Relational DBs are very mature and robust.

(Also, if you ARE going to go with a relational DB, may I suggest PostgreSQL).

Regarding the questions you've asked, the answer depends greatly on the NoSQL database you're going to choose. Lots of best practices related to big-data databases don't necessary use JOIN operations (at least not to the same extent as relational dbs) and this is mainly because when it comes to distributed databases, storage is cheap but JOINs on huge datasets in really really expensive.