>So, if we want to make performance testing, we need to compare only
>InterClient vs. type 4 JDBC driver, because we do not compare DBMS,
>but driver implementation. Also, for each test fresh database must be
>used because of garbage collection. And definitely logging must be
>off. :)

I'm going to rerun the tests doing just that - comparing the different
drivers. Of course, do we know that InterClient was fast in the first
place? We can at least make sure we are not slack in comparison to
InterClient.

I'm also going to run the tests against Postgres again, but I'm going to run
all queries in the same transaction, with autocommit turned off. I would
also like to be sure logging is turned off (it should be, I have the command
line switch in there, but...)

Richard


_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com