SQL Server 2008 Full Text Search (FTS) versus Lucene.NET -

I know that in the past SQL 2005 vs Lusen There is a question on the net, but came out since 2008 and they did a lot to change it and thought that someone could give me a professional / opposition (or link to any article). I built a medium-sized knowledge base on the top of FSS of SQL Server 2005 (probably in 2006) (probably:

2 GB of indexed text), and now it has moved to IFTS 2008. Both situations have worked well for me, but the move from 2005 to 2008 is actually an improvement for me.

My situation was not like the stack overflow that I was indexing the data which was only fresh at night, though I could join search results from multiple content statements in each other and in relational tables. Was trying.

In 2005 FTS, it meant that to include each, the search must be done on the index, the whole result and then the DB engine incorporates those results into the relational tables (this is for me all Was transparent, but it was happening and was expensive for questions). The 2008 IFTS has improved this situation because database integration allows many contingible results to be part of a query plan, which has made many searches more efficient.

I think both the 2005 and 2008 FTS engines, as well as Lucene. Net is an architectural business which is going to align better or worse in a lot of project conditions - I just got lucky that the upgrade worked in my favor. I can see completely why the 2008 IFTS will not work in the same configuration as the 2005 Stack Overflow.com, for the use of such cases as the OLTP nature of excessive use. However, I will not leave the possibility that 2008 IFTS can be separated from the heavily-loaded transaction load ... but it also seems to be enough to accomplish this as much as Lucene.NET ... and quiet The factor of Lucene.NET is difficult to ignore;)

Still, for me, the ease and functionality of SQL 2008's IFTS in most situations might be perhaps the 'cool' factor of Lucene (though it's easy to use Do i Sector has never been used in the production system, so I'm reserving comment on that).

I want to know how much more efficient is Lusen in stover overflow or similar situations (now it has been implemented?).