Manticore as replacement for MySQL database

[d47081]

Hello,

I think about MySQL + Sphinx replacement to Manticore Search.

So few questions:

1. Is the data permanently stored on disk or it is just store indexes in RAM like Sphinx do?
2. Does provide Manticore project tools like MySQL Workbench with it UML diagrams + sync / export tools?

In the other words, can I import MySQL DB project to manticore and drop MySQL and Sphinx solutions at all?

I know it's just sphinx fork, but for me beginner, it looks like complete DB for the search engine.

[sanikolaev]

Is the data permanently stored on disk or it is just store indexes in RAM like Sphinx do?

Both Sphinx and Manticore support binary logging. They also behave similarly (at least when comparing Manticore with the open-source version of Sphinx, I'm not sure about the latest versions) in terms of RAM chunks and disk chunks: documents are first accumulated in a RAM chunk (which is covered by the binlog and also flushes to disk regularly), then they are flushed to a disk chunk.

Does provide Manticore project tools like MySQL Workbench with it UML diagrams + sync / export tools?

I don't think so. There's an open issue about adding support for MySQL Workbench - #1279.

I know it's just sphinx fork, but for me beginner, it looks like complete DB for the search engine.

That was one of the main ideas of the fork: to transition from a full-text index/filter to a full-fledged database. That's why we:

• Added docstore with all fields stored by default
• Added RT mode by default
• Renamed "indexes" to "tables"
• Added backup tools including mysqldump support
• and made many other minor improvements.

In the other words, can I import MySQL DB project to manticore and drop MySQL and Sphinx solutions at all?

Be aware that Manticore is not fully ACID-compliant. For example, there's an open issue about REPLACE atomicity - #397. Transaction support has been improved slightly since forking from Sphinx, but not significantly. If this isn't a problem for you, then yes, you can give it a try. Just make sure to back up regularly (it's best to do a logical backup with mysqldump in this case), which makes sense with any database.

[d47081]

Thank you for explained reply

So manticore server (beside RAM pull) keeps data on disk storage and does not require third-party source DB?

I doubt, because of sphinx mostly using to make keywords + document ID relations, then usually request passing to DB to extract more info by document ID. Even it's possible to store text fields in index, that's not good idea there (in sphinx), as MySQL server more resistant to server faults and have build-in solutions to restore the data without mysqldump.

Asking because Sphinx is called Search server (operation software) and Manticore is Search database (storage software) so I don't understand yet, maybe need more time to read the docs and play with local instance.

Or, maybe it's just MySQL-like abstraction environment for the Sphinx.

[d47081]

Just found reply to my question:

yes, it could be used as the permanent storage and simple object-oriented replacement for MySQL

I've found for myself following library that describes project logic in few words in php:
https://github.com/manticoresoftware/manticoresearch-php#getting-started

Plus one file for Sphinx users:
/etc/manticoresearch/manticore.conf

Not sure about existing project and data migration to the manticore models, foreign keys, transactions, etc but for search project that writing 'from zero' - it's pretty cool solution and not require additional DB on back-end.