Re: Oracle memory allocation on Linux 2.6

From: Robert Klemme <shortcutter_at_googlemail.com>
Date: Fri, 4 Apr 2008 01:00:32 -0700 (PDT)
Message-ID: <1ba6ff7f-8b0a-4998-9723-45ea3efed2d2@a22g2000hsc.googlegroups.com>


On Apr 3, 3:36 pm, Mladen Gogala <mgog..._at_yahoo.com> wrote:
> On Thu, 03 Apr 2008 06:17:23 -0700, Robert Klemme wrote:
> > The point where I disagree is when you turn down improvements in
> > management automation.
>
> Management automation? I know few managers that could be replaced by
> a Perl script, but I wasn't aware of the trend?
>
> > Basically systems become more complex all the
> > time and we also learn more about system behavior. It is much more
> > efficient to try to put that knowledge in code than to train a lot of
> > people to do the same.
>
> HAL 9000 awaits us. How do you put knowledge in the code?

We do that all the time when developing software.

> How do you
> get a computer to fix a horribly botched applications in which queries
> are created on views upon views upon views?

Of course you don't. As I said in my earlier posting, I'm all in for people that know their job and do good work. But this thread is about automated memory management and you seem to generally dislike features like this while I tried to point out that they can actually work by pointing at the JVM example. The fact (?) that Linux MM is not as good as it could be does not prove that automation like this is bad in general.

> > The Java Virtual Machines of today are an
> > extremely good example of where this can lead - performance has much
> > improved over earlier versions and this is because the JVM is
> > "intelligent" enough to do optimizations on the running code.
>
> Hmmm, I am not sure that JVM can outrun Perl.

I am sure you can find benchmarks that prove each claim - Perl being faster and Java being faster.

Cheers

robert Received on Fri Apr 04 2008 - 03:00:32 CDT

Original text of this message