Keyword search pattern to relational data is the most important and the highlighted area within a search and information retrieval community. For the system evaluations, there are many approaches followed but there is a lack of standardization. The result of lack of standardization affects performance of the system. The number of queries completed successfully in a query workload is performance wise not showing good results for relational keyword search system. The solution to above problem is to develop a novel technique for efficient document retrieval using relational keyword search system. The new system is developed to manage uploading and downloading of data from disk to improve performance and reuse dataset and query workload to provide greater consistency of results. A scalable document retrieval improves the search performance in terms of execution time, cost efficiency and apply ranking to the document depends on query weight. The new system gives 30% to 40% reduction in the search execution time compared to the existing system.