Can I use Elasticsearch for query DB in replace of standard DB?

MongoDB, for example, is great but it has no full-text search (Atlas Search does, but its a different thing). Also you need to explicitly index fields.

Elastic allows dynamic indexing per every field in the document. So basically every field is indexed and you can query against all of your data very fast (btw is it good to do so? just drop tons of documents and let them all be indexed?)

So I guess that when I need just to query data, without any mutations or ACID transactions - for example - give me the list of employees who work for a company StackOverflow, then add filter of who works for more than 5 years in role of developer, then add full-text search for name starts with "Dan". All of these can be achieved by Elastic.

So probably it can be a full replacer for MongoDB/MySQL/Postgres/etc - for query only. Isn't it?

Yes, but still there are something you need take into consideration.

1. elasticsearch supports join , parent-child query with very poor performance comparing to mysql.
2. elasticsearch query data slower than other db, cause of boost query.
3. Pay attention to aggr, such as count, distinct, there is a precision problem. As elasticsearch is designed with shards, aggregation from mutilple shards will be slow and inaccuracy.
4. Proficiency of elasticsearch query counts, as elasticsearch DSL is more complex than traditional SQL.
5. elasticsearch use more disk than other db, cause of analyzation of text field.

Query performance counts for every kinds of db.