cancel
Showing results for 
Search instead for 
Did you mean: 

Oracle tips and best practices.

Amy_Young-Leith
Star Contributor
Star Contributor

With almost 200 million documents in our repository since 2016, and growing each day, I am encountering performance difficulties on custom SQL queries used to create data providers in Dashboards. 

 

I'm working with an Oracle DBA at my org, who is not well versed in OnBase, and he is looking at some of my queries to determine if additional indexes might help, etc. I'm an avid SQL query writer, and just venturing into more database level performance tools and approaches. I just write my query and it works... but takes a long time. Most of my queries are time limited, often to a 13 month lookback.  🙂 

 

I was wondering if it's normal or helpful to create a view for each document type group (which corresponds for us to a business unit implementation), and create a view for the "average document." Basically the big join of all keywords on those documents? 

What other good tools, tips or tricks are out there in the real world. I have read at length both the Database Reporting Guide and the Database Reference guide for our version, 22.1. 

6 REPLIES 6

Yep, provided those already. I'll point out more clearly the spots you reference. Thanks. 

Thomas_Reu
Elite Collaborator
Elite Collaborator

another thing you can do is a nightly/weekly backup to a reporting database that is held on a different database than your onbase prod database.  Optimally, on a different server.  Then point your reports to the views you create there.  Since this db is an independent db, hopefully on a separate server, and not your onbase db you can make whatever modifications (indexes, etc) are required without getting any approval from anyone.   Also, if you have issues creating efficient queries the onbase prod server won't be affected since it is on a completely separate server.

 

Granted you may have to rerun any index scripts (etc...) after the backup/restore.  But, it might be worth it.  You'll have to test and make that call yourself.

 

Note: the above is irrespective of the database type.... 

I have a warehouse like this for our Banner database. Users defined what information they need then each night I build a subset of that data for reporting the next day. 99% of our student data is OK to be 24hrs hold.