Re: Running Instances on Windoz

vslabs_at_onwe.co.za (Billy Verreynne) wrote in message news:<1a75df45.0311250241.37874a2_at_posting.google.com>...
> joel-garry_at_home.com (Joel Garry) wrote
>
> > Well, I see _every_ IT shop that has Windows, mission-critical or not,
> > spending inordinate amounts of time dealing with the basic security
> > issues that affect all Windows software, for reasons that have been
> > obvious for years. The success is in spite of, not because of.
>
> Agree fully. Security is an issue. A Windows out-of-the-box
> installation is not secure and needs to be hardened.
>
> Back in the 90's one of my colleagues went on about NT's C2 security
> certification and what not. I hacked his NT box in about five minutes
> (using a SQL*Server 4.5 backdoor) and took over his NT sysadmin
> account. He was not impressed with either me or Microsoft. :-)

http://catless.ncl.ac.uk/Risks/21.24.html#subj10.1

>
> The real problem is that security is Microsoft development and
> architecture was not a core consideration back then. Now they are
> attempting to make it one.. a tad too late in many respects I would
> think.

To me that is so bizzare. How could Cutler & Co. make such a mistake?  They of all people knew better.

>
> > > That depends on how well you know Windows, what hardware you are
> > > running on it, and how you configure it.
> >
> > I claim not to know it well at all. But I shouldn't have to. The
> > problem there is the people who are charged with knowing are not given
> > the tools to really find out what any problems are. That's considered
> > a feature, I think.
>
> IMO one of the issues is the perception that because you have all the
> GUI tools on Windows (and there's a plethora of them), you do not need
> to know *what* those tools exactly do and *how*.

Perception is the issue.

>
> IOW, these GUI tools turn Mr Sysadm/DBA into an ignoramus. On Unix,
> the effect is usually the opposite due to the lack of GUI tools.

Oracle has managed to undo that with OEM. :-O

>
> > So tell me, how does one configure Win2000 with 9iAS (all three
> > tiers)?
>
> The same as on Unix/Linux... with difficulty. Hell, I did manage
> (after a while) to get 9iAS working on vanilla RedHat 9 after
> borrowing from SO's from 9i Enterprise Server... :-)

Well, I was going for a FAQ answer on something supported, but what the hey, you always have more entertaining answers! :-)

>
> > How competent do you have to be to say that basic functionality like
> > starting up and shutting down the database with the box in a
> > reasonable amount of time cannot be made to work with Oracle? There's
> > plenty of competent anti-windows expertise out there.
>
> As I said Joel, Windows is not Unix. And it only requires a few
> minutes of reading the Win32 API specs (the Microsoft version of Unix
> man pages IMO) to understand how services work and what the
> requirements are.

So why do we have issues of Oracle closing bugs with "Not feasible to implement?" Why does support recommend using batch files to start and stop? I don't think a few minutes of reading specs answers this at all.

>
> BTW, it is much easier to configure Oracle startup/shutdown for a
> Windows box than a Unix box in my experience. The only downside on the
> Windows platform is the lack of scripting like we have with Unix...

That is not my experience. Once you throw 9iAS into the mix, you have timing issues up the ying-yang. I am very bothered by scripted boots that do not work the same each time. What do you mean by lack of scripting? Isn't that what that console group policy http://www.cs.tufts.edu/~kisrael/kingmissle.html#penis or whatever the heck it's called is for?

>
> > > > And if you
> > > > don't think bg intends it to take over from unix, what part of deepest
> > > > darkest Africa have you been living in?
> > >
> > > The arse side.
> >
> > I better not ask what they put in the coffee there!
>
> Hehehe... The country on the arse side of Africa, i.e. down under in
> the south, is South Africa. And down here we do put milk and sugar in
> coffee.. ;-)
>
> > I think it is another version of VMS, myself. But with a front end,
> > support infrastructure, and general corporate attitude that masks the
> > good thing that that could be.
>
> Tend to agree with that. If anything, Microsoft made NT look too much
> like Windows 3.1 IMO. And from there the perception grew that NT is
> indeed like Windows 3.1.
>
> Not that I'm a die-hard Windows NT/2000/XP fan. If anything my
> favourite o/s still is Siemens' BS2000.. but then that's probably just
> nostalgia.. ;-)

An honest OS - explicitly BS! :-)

>
> > Whoopie. I've seen Unify truck on in unix environments that Oracle
> > would look like a boiled goldfish in. I'd be much more interested in
> > a story like yours that s/OS\/2/unix/g.
>
> Well, never tried NT's POSIX subsystem. Have however done some good
> stuff using the BS2000 POSIX subsystem. :-)
>
> I think the real question you are asking is if Oracle and Windows are
> suited for a good and stable marriage in the corporate environment?
>
> Cannot really give an opinion on that.. except to say that Unix and
> Oracle in my experience is a techie match in heaven.
>
> However that does not say that Windows cannot do the job when it comes
> to that type of environment. Besides, the deciding factor about what
> technology to use is not which one results in us wearing the kewlest
> Raybans, but which one fits the business environment, requirements and
> strategy the best.
>
> Our job is to make it work and work well. I'm sure that like all
> techies you also thrive on challenges like that. :-)

Well, I'd still like to see someone honestly compare Windows and Linux. Particularly some replicable ordinary hardware configuration that runs good on Linux and crappy on Windows.

jg

--
@home.com is bogus.
RSTS rulez!