Performance evaluation of searching using various indexing techniques in Lucene with Relational Databases

The Organizations commonly use relational databases for transaction processing, but big portion of database operations involve select operation. As data grows beyond few million records selection tends to take much time in whole transaction. One approach is to build indexes in database on columns which are frequently used in selection. If there are more than one table (which is general case) selection takes more time. Another approach is to use searching framework for searching records. Apache lucene is very popular, fast open source searching framework used in many projects. So here we are trying to evaluate use of lucene searching to find records fast so as to get performance benefits from lucene's fast searching capabilities and offload selection work from databases. We will evaluate different indexing types in lucene to see which best fits to our need. At last we evaluate that is this arrangement can provide performance benefits, and which index type is best suited for that.