Does anyone have experience using Apache Lucene as their primary data provider?

Question

Java

Hello.
We screwed Apache Lucene to the application. We use it for full-text search on the system. You know, it's like that input at the top "Find a question, answer, tag or user". And then the question arose, is it possible to use Apache Lucene as the main data source? Those. do not go to the database for data, from the word at all. An application from the field of "bloody enterprise" - SED. There is more and more data, the samples are getting slower. It is clear that it is possible to optimize queries, optimize the DBMS itself. But after all, Lucene does a good job of quickly accessing data according to specified criteria.
Actually, I would like to hear advice on using (or not using) Lucene, or a similar engine, as the main source of application data.
Thank you for attention.

Answer 1

Had experience with elasticsearch, but not as a primary repository.
The adjacent team used it as the main one, then from their words (for what I bought, for that I sell).
The experience was rather negative. There were problems with reliability, backups, etc. The data is denormalized, if something goes wrong, it will be a big problem to correct it. A couple of times the indexes broke. In the case of a traditional RDBMS as the main repository, one would simply have to run a reindex. And here is a complete, excuse me, ass. They don't do that anymore.

Answer 2

I won’t give proofs, but it seems to be a common belief that using Lucy as the main storage is a way to shoot yourself in the foot