May 26th, 2010, 11:41 AM
Spring batch performance on large volume
We starded a new project that consist to modernize and write an application replacing COBOL batch. The target platform is mainframe zOS/DB2 Database or Unix IBM Aix/Oracle Database.
The existing application treats 10 million operation per day (importing flat file in database, executing business operations on this database and finally producing a flat file). Is there some references about performance using spring batch in this case (recommended hardware configuration, best practices..)
May 26th, 2010, 11:27 PM
I've worked on a similar project for a client a couple of years ago. It was a very similar situation, with Spring Batch replacing an old COBOL system, but still actually running the jobs on the mainframe for various reasons. If I'm not mistaken it was using DB2 as well. I don't know what the volumes ended up being, but it was in the millions and I believe the project went live fairly recently without issues. However, I'm under NDA unfortunately, so I can't give much more details than that. However, 10 millions records in a day isn't really that large honestly. A simple java batch solution shouldn't have any troubles with that at all. Personally, I've seen Java batch solutions handle more than 100 million records a day and still beat the performance estimates. Sorry I can't give you any case studies, etc though.