Re: Sniffing redo logs to maintain cache consistency?

Following up on Andrej Gabara, 25 Feb 2003:

> have connection pooling and prepared-statement caching for improved
> performance as well. And most of our JDBC calls make use of prepared
> or callable statements, using bind variables.

You are halfway there already.

> and performance would be better. But effective caching is difficult to
> do when you have business logic in the database, unless you can push
> change events to the cache quickly. But then, maybe the lack of
> effective caching isn't our biggest problem, we need to investigate more
> time into this.

Yes, definitely. Your problem appears to be that you map the Java objects to the PL/SQL calls almost 1-1. So, when an event somewhere in one screen causes a change in the object, you get a call to the db even if it is just to flip a bit. What you have to do is look at decoupling your objects from the actual tables. The technique described in the DAO pattern is quite good. Basically, a method to change a bit on an object does not need to be immediately serialized.

>
> Maybe I should be emberassed for asking the "sniffing log" question
> in the first place, but I think I had a good reason for it, so I'm not :)

You should not be embarassed at all.

-- 
Cheers
Nuno Souto
nsouto_at_optusnet.com.au.nospam