I have a database that stores details of Code Chekins from various SCRs. One of the table in this database store Commit Comments for each checkin. I am trying to develop a search feature which with the help of Postgres posix notation searches through this table trying to match a regular expression on this comment field and return all the matched.
I have already got this to work, but the main problem here is the performance of this search. For a fairly big database it almost takes 15-20 mins for a search to complete and as its a web frontend waiting for the result this is totally unacceptable time for a medium sized database. I figured that creating an index on this text field might help but I am unable to create a btree index because data for some of the rows is too big for potgres to create index on it.
Is there any other solution to this? Are there any other indexes that can be created which again should not be language dependent?