# Regarding the operation mechanism of full text search engine

**URL:** <https://forums.percona.com/t/regarding-the-operation-mechanism-of-full-text-search-engine/1088>\
**Category:** Other MySQL® Questions\
**Created:** [February 15, 2009, 9:56pm UTC](https://forums.percona.com/t/regarding-the-operation-mechanism-of-full-text-search-engine/1088 "2009-02-15T21:56:08Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![csanghvi](https://avatars.discourse-cdn.com/v4/letter/c/3bc359/32.png) [@csanghvi](https://forums.percona.com/u/csanghvi)\
**Post date:** [February 15, 2009, 9:56pm UTC](https://forums.percona.com/t/regarding-the-operation-mechanism-of-full-text-search-engine/1088/1 "2009-02-15T21:56:08Z")

</div>

Can anyone here please explain me how myisam performs a full text search. I read the reference manual and i know about how mysql parses the text or varchar fields marked as full text. But it doesnot explain how mysql indexes these full text words. Are these words stored in the key buffer and then the search on them is also based on B-tree like on any other index?  
Please help me understand the internals of full text search engine.  
When benchmarking the performance of my application, the speed for full text search on a table with 5billion rows was substantially better than the speed of search by using primary key. I am baffled as to why should this be?  
Why is it that with increase in the database size the performance in speed of searches reduces?

---

<div class="post-metadata">

**Author:** ![xaprb](https://avatars.discourse-cdn.com/v4/letter/x/49beb7/32.png) [@xaprb](https://forums.percona.com/u/xaprb)\
**Post date:** [December 25, 2009, 7:24am UTC](https://forums.percona.com/t/regarding-the-operation-mechanism-of-full-text-search-engine/1088/2 "2009-12-25T07:24:06Z")

</div>

Larger data always means slower performance, especially as the data gets too big to fit into memory anymore. Full-text is built as two B-trees as I recall.