I have two exactly the same tables with 4.5 Million resumes indexed with FTI, and a clustered unique ID.
One table has the standard Microsoft 154 English stopwords.
The other table has 10000 stopwords.
Prior to running each query, i ran an
UPDATE STATISTICS Tablename WITH FULLSCAN
Then i ran a DBCC Freeproccache.
Then i ran the following query:
set statistics time on
--select count(*) from Profiles91313bb
select count(*) from Profiles
where contains(doccontent, 'resume')
and the results are basically identical, around 1.05 seconds.
Is this because 4.5 Million records, with up to 10000 stoplist words per resume, is just not enough data to cause index bloat ?