Mar 30th, 2011, 03:41 PM
SpringBatch on DB2 z/os
We plan to use Spring batch at work for our high volume batch processing. I have a few questions, would appreciate your input
1) How do you plan your spring batch meta tables. Do you create those tables in a centralized DB to be used by all batch apps or create independent tables for each batch
2) Can having the batch app access spring tables in one type of DB (say DB2) and application related data on another (say Oracle) affect the performance of the batch. Or do both need to be on the same DB server to improve performance.
3) We ideally plan to put spring meta tables on DB2 on z/OS environment. I see there is no specific DDL provided in spring batch jar for DB2 on z/OS. I tried a POC which forced me to change BIGINT to INTEGER and when creating tables on DB2 on z/OS
Did anyone have any experience putting these tables on DB2 on z/OS. Any issues or performance impacts to be know before hand if we go this way
Thanks much for your help
Mar 31st, 2011, 01:02 AM
your life will be easier if the batch metadata are located in the same database as the business data: better guarantees (regarding number of skipped items, etc), easier to set up and faster (no XA, you can use local transactions).
Mar 31st, 2011, 10:13 AM
Thanks for your input. But doesn't that mean a lot of overhead on each and every DB wherever you plan use spring batch. Also my question on DB2, is it ok to change BIGINT to INTEGER. Any experiences down the lane..
Mar 31st, 2011, 10:50 AM
that's the price to pay for SLA (number of skipped items, restart, and so on). Sorry, I can't help you on DB2.