Re: metalink still unuseable the 2nd day ...

From: Mladen Gogala <>
Date: Thu, 12 Nov 2009 13:50:14 +0000 (UTC)
Message-ID: <hdh3qm$ne3$>

On Wed, 11 Nov 2009 19:51:01 -0800, Noons wrote:

> On Nov 12, 11:26 am, joel garry <> wrote:

>> > problems with it.  It'll go away, I'm sure.  But did it really need
>> > to be like this?
>> I'm leaning towards "yes, meatlink was so creaky and fungusoid, it
>> did."  And yes, I'm one of the loudest squawkers about it.  But this
>> too will pass, just like the Sidekick fiasco, and Audi sudden
>> acceleration, and hacking of Apple keyboards.  Oh wait, that last one
>> just started.

> We disagree here. I didn't have the slightest problem with old
> Metalink.
> It did its job quite well, metaphorical adjectives apart.
> Of course: it didn't support OCM. But hey: I never asked for OCM
> anyway,
> neither did a lot of users. It's an imposition from Oracle: not needed,
> not asked for. To cause this mayhem and bad blood with clients for the
> single reason of updating something that works so that something else no
> one asked for could be supported, is sheer madness.
> Dunno about you, but after spending weeks disabling half the information
> about my systems that OCM wanted to collect and settling finally on a
> "disconnected" setup, I really don't see much value in what it does that
> wasn't already addressed with SR templates.
> And no: there is NO WAY anyone from Oracle Support will EVER convince me
> they need to know the IP AND the MAC address of every network interface
> in my servers in order to "collect information" about my databases.
> That one is bordering on lunacy and just about breaks every mantra of
> basic system security! Yes, it collects and sends MAC addresses - by
> default!!!
>> The oracle-l thread about quotes from support people on the phone
>> trying to access MOS/metalink was pretty funny, too.

> Oh yes. Not sure how close to reality it was, but funny indeed!

A few years ago, huge shift in philosophy happened in Oracle Corp. They stopped publishing internal information, 11G has a whole slew of the new processes that are completely undocumented, they started producing features as options, my pet peeve being the AWR report, and their prices went up. Also, flashback process added a whole new class of events and processes which are completely undocumented and not tunable. They also started buying their competitors: Peoplesoft, BEA and, the latest, SUN Microsystems. Their licensing policy also used to be pretty lenient. With OCM, they are engaging in "big brother is watching you" stuff which I definitely do no like. At the same time, the code quality went down rather drastically: OEM was practically unusable before 11.2 are there are plenty of rather critical bugs. The latest thing I finished testing was de-duplication and compression of LOB columns. I had 24GB expdp, produced by the version, file which I tried importing into 50GB of space. One would expect that to succeed, with the de-duplication turned on and compression set to "high", but it ran out of space. So much for the "securefile" storage.
Basically, this is not the same company as it was when I started doing Oracle 4.1.4 on IBM PC XT and when the Oracle installation set came in 3 gray boxes with 5 360k 5.25" floppies. At that time, Oracle was open, helpful and nimble company that was fighting IBM, Ashton-Tate, Sybase and Ingres. Now, they are a monopolistic juggernaut with a choking stronghold over the database market, prince charming turned into the king of Id.

Received on Thu Nov 12 2009 - 07:50:14 CST

Original text of this message