Re: Benchmarks - Java vs PL/SQL

You are absolutely correct. So, for those several thousand applications out there that spend most of the week querying DUAL, you'd better use PL/SQL.

Richard
rvsutherland_at_yahoo.com

<jdarrah_co_at_my-deja.com> wrote in message news:920l21$49$1_at_nnrp1.deja.com...
> I agree that its important to write effiecient SQL but for applications
> that are transaction and query intensive, bad data access will kill
> performance. In the test I conducted 10,000 iterations of "select 'X'
> from dual" took 7.5 sec in a java stored procedure vs .15 in PL/SQL. I
> doubt even the sql savvy technician could hint the above statement
> enough to put java stored procs over the top.
>
> In article <t47hfvhs05nr80_at_corp.supernews.com>,
> "Richard Sutherland" <rsutherland_at_In-Touch.net> wrote:
> > Actually, the "data access advantage" has little to do with this whole
> > subject. What REALLY matters is the SQL written. Lousy SQL sent to
 the
> > database in 0 nanoseconds is still going to lose to the savvy SQL
> > technician.
> >
> > Richard Sutherland
> > rvsutherland_at_yahoo.com
> >
> > <jdarrah_co_at_my-deja.com> wrote in message
> > news:91vtfc$cfs$1_at_nnrp1.deja.com...
> > > The artical you mention glosses over one very important point. To
> > > quote Larry:
> > > "It was not the goal of these tests to examine any specific data-
 access
> > > API or transport which may offer an "data access advantage" to one
> > > specific language platform over another (e.g. JDK 1.2.2 with a
 native
> > > JDBC driver might be able to access the database more efficiently
 than
> > > JDK 1.1.7 using an ODBC-bridge)."
> > >
> > > The primary purpose of most database applications that I have been
> > > involved with is to select information from the database, display
 it to
> > > a user, and allow that user to manipulate the data. In cases like
> > > this, the most important performance metric IS the data-access API.
> > > Larry's performance tests miss the forest for the trees. I have
 little
> > > doubt that java can run circles around pl/sql with number crunching
 and
> > > string manipulation tasks but that won't really matter if the
> > > application is spending 99% of the time waiting on the jdbc layer to
> > > return the data to be crunched.
> > >
> > > In article <3A42AAA3.3020009_at_interealm.com>,
> > > Roby Sherman <rxsherm_at_interealm.com> wrote:
> > > > Avi Abrami wrote:
> > > >
> > > > > I am looking for some performance
> > > > > measurements/comparisons for stored
> > > > > procedures in PL/SQL versus stored
> > > > > procedures in Java.
> > > > >
> > > > > We are trying to decide whether it is
> > > > > better to write our stored procedures
> > > > > in Java or PL/SQL.
> > > > >
> > > > > TIA,
> > > > > Avi.
> > > > >
> > > > >
> > > > Try http://www.orageeks.com/technotes/larry/languages.html
> > > >
> > > > --Roby
> > > >
> > > >
> > >
> > >
> > > Sent via Deja.com
> > > http://www.deja.com/
> >
> >
>
>
> Sent via Deja.com
> http://www.deja.com/
Received on Fri Dec 22 2000 - 19:56:32 CST