Re: Java and Relations DBs..

["Larry Loen" <lwloen@xxxxxxxxxx>]

>So I guess what I'm asking is, will it hurt anything if I do
>write my own pooling procedures and the driver I'm using
>already impements pooling?

Not on OS/400.  I still code pretty much to 1.1.8 interfaces, but a
colleague down the hall said that you could rely on the data source
improvements added in JDBC 2.0 (I think that came in around Java 1.2).
That will end up working its way back to WebSphere (and you'll have to
master a bit of the config/admin GUI) but your code will be portable (and
being 1.2+ dependent nowadays is what many are doing for other reasons).

Either way, you're ultimately just passing around references to the
connection object, so you can certainly roll your own or rely on data
source stuff.  (I wouldn't mix and match the approaches, since WebSphere
caches stuff for you automatically, and there might be some side effect I
can't imagine just now, but most people wouldn't really want to do that.
Just do one or the other).

Anyway, doing some form of connection pooling is near the top of what we
recommend for performance tuning.  For any serious performance in SQL, it
is kind of something that you're expected to do.  The virtues are widely
advertised, so I won't rehash them here.



Larry W. Loen  -   Senior Linux, Java, and iSeries Performance Analyst
                          Dept HP4, Rochester MN