Indexing system vs raw DMBS connection

I have created a new dataset that has 11,000,000+ rows with 4 pivot tables in MySQL. The tables are not that deep, only 6 - 12 cols per. My question is I've set up the Lucene Apache SOLR to index the data, and it work's great for searching .. BUT I haven't noticed a dramatic load time in searches. If I run a raw SQL command, the difference between the two is negligible. At what point is it more beneficial to use a indexing engine vs scripting a raw query yourself? Now granted I have the ability to set "weight" and all that via SOLR .. But my impression was that I would have dramatically reduced overhead on the server ... Is it just that my dataset isn't large enough / complicated enough to illicit these finding? Other than the obvious advantages being weight, rank and sort/filter -- What is the overhead advantage of using a indexing system such as SOLR?