I have a large web application with a complex and performance intensive search function. The search allows the user to select and filter on any of 250+ columns spread over 20 or so tables in the Oracle database. The initial implementation uses a fairly conventional approach using Hibernate and its criterion API to build up the query dynamically based on the users selection, and then run that query against the main system-of-record database. It works, but I am not confident that it will stand up to the eventual production load with many thousands of concurrent users. Also this new search load will impact other users of the main Oracle DB - so I am looking for other implementations that will perform fast, scale, and not adversly impact the other DB applications.
We reseached and ruled out Lucene based solutions. A NoSQL document based solution like MongoDB would work - but this involves more work and challenge in syncing the Oracle data to and querying from the MongoDB and its json format.
SQLFire seems appealing in that I should be able to easily replicate my Oracle data structures in the SQLFire DB, and use the same Hibernate implementation to query against it. So my questions to the SQLFire team are:
- Is this a good use case for SQLFire?
- Are there any handy tools/techniques for syncing the main Oracle data with the SQLFire DB? As the main DB gets updated the search DB would also need to be updated in near-real-time.
thanks!
Jeff