Oracle Database Discussions

Dear Oracle Forum:

We have a bit of controversy around the office and I was hoping we could get some expert input to get us on the right track.

For the purposes of this discussion, we have two machines, development and production. Currently, on each machine, we have one database with multiple schemas, say, one for sales data and another for inventory. The sales data has maybe 200 tables and the inventory has another 50. About 12 times a year, once a month, we have a release and move code from dev to prod. The database is accessed by several hundred Pro*C and Pro*Cobol programs for online transaction processing.

The problem comes up when we need to have multiple development environments. If I need to work on something for May that requires the customer address field to be 50 characters and somebody else is working on something for July that requires the customer address field to be 100 characters, we can’t both function in the same schema. We have a method of configuring running programs to attach to a given schema/database. Currently, everything connects to the same place. We were told that we should not have the programs running as the owners of the schemas for some reason so we set up additional users. The SALES schema is accessed with the connect string: SALES_USER/SALES_PASS@maindatabase.world. (I don’t know where we got dot world from but that is not the current discussion.)

One of the guys said that we should have 12 copies of the database running, which is kind of painful to think about in my opinion. Oracle is not a lightweight product and there are any number of ancillary processes that would have to be duplicated 12 times.

My recommendation is that we have 12 schemas each for sales and inventory with 12 users each to access them. We would have something like JAN_SALES_USER, FEB_SALES_USER, etc. Each user would have synonyms set up for each of the tables it is interested in. When my program connects as MAY_SALES_USER, I could select from the customer table and I would get my 50 character address field. When the other user connects as JUL_SALES_USER, he would get his 100 character address field. Both of us would not know anything different.

Another idea that came up is to have a logon trigger that would set the current schema for that user to the appropriate base schema. When JUL_SALES_USER logs in, the current schema would be set to JUL_SALES, etc. This would simplify things by allowing us to avoid having something like 2400 synonyms to maintain (which could be automated without too much difficulty) but it would complicate things by requiring a trigger.

There are probably other ways to go about this we have not considered as yet. Any input you can give will be appreciated.

Regards,
/Bob Bryan