I've searched on multiple sites, documentation, metalink, google... But have not come up with anything to resolve an issue with a production database.
We use a third party application, which was written such that it uses dblinks to access tables between schemas. They say it was written that way so that it could be implemeneted with multiple servers.
We have them in one instance, three schemas. Can't change their code.
We've tried, bumping up the buffer cache, setting sdu/tdu, loopback (which worked - but caused a log file from the application to fill with their message.) There has been no user perceived improvements in response time and the buffer cache ratio gyrates between 10 and 90% (almost the same curves as logical/physical reads, and sqlnet/dblink to and from client.)
I believe the buffer cache ratio is bad because almost every query 'thinks' it doesn't have the data locally. I can't support my theory with documented facts.
Any Ideas?
We use a third party application, which was written such that it uses dblinks to access tables between schemas. They say it was written that way so that it could be implemeneted with multiple servers.
We have them in one instance, three schemas. Can't change their code.
We've tried, bumping up the buffer cache, setting sdu/tdu, loopback (which worked - but caused a log file from the application to fill with their message.) There has been no user perceived improvements in response time and the buffer cache ratio gyrates between 10 and 90% (almost the same curves as logical/physical reads, and sqlnet/dblink to and from client.)
I believe the buffer cache ratio is bad because almost every query 'thinks' it doesn't have the data locally. I can't support my theory with documented facts.
Any Ideas?