Re: Question about entry level Oracle DBA

Daniel Morgan <dmorgan_at_exesolutions.com> wrote in message news:<3DAC31C9.DC34EBA2_at_exesolutions.com>...
> Responses interspersed below:
>
> Joel Garry wrote:
>
> > Daniel Morgan <dmorgan_at_exesolutions.com> wrote in message news:<3DAAE348.D6B12ADA_at_exesolutions.com>...
> > > los wrote:
> > >
> > > > If every job listing is asking for 3 to 5 years of experience in Oracle,
> > > > how are entry levels supposed to break into the business? Is it impossible
> > > > to get an entry level Oracle position right now?
> > >
> > > You are supposed to become a developer and learn about the architecture, how
> > > to code, how to tune, how to debug, etc.
> > >
> >
> > Who sez? That would be for a development DBA, does not necessarily
> > generalize to all DBA's (except for those morons who post things
> > like "Must have 5 years experience with O9R2").
>
> I sez! Because if all I want from a DBA, and I interview and hire them for my clients, was someone that
> could do install, back-up and restore I would just send the UNIX SysAdmins to a one week class and save my
> clients the money. If you can't write SQL and PL/SQL you can't tune it. Your opinion in a code review will
> be worhtless. You won't know the impact difference between a static cursor and a dynamic cursor in a
> multiuser environment. Asking for advice on whether to use an IOT will be greeted with a blank stare. And
> most importantly ... when someone wants access to utl_file the DBA won't have any rational reason upon
> which to determine whether such access is appropriate.

Well, you sez wrong. Not necessarily for your particular situation (and perhaps for what I call development DBA's), but in the larger world there are more situations. Now, take me. There was a time I was a SQL coder, but even then, I didn't like it, because I already had 10 years of experience with R's using other languages - and SQL in the '80s was by no means something that appeared to be long lasting. Now... I have to think about it. PL/SQL? I passed the test. Whenever I've run into it in a production environment, it's susually been one of two situations: a programmer doesn't understand why his code doesn't work, and I've usually been able to hack through that; or it is "required" for some process to do something for the db, and there is usually
some doco that says how to do that. Nowadays, I have Tom's book, but I haven't had to crack it yet... but note that I have found times when the O-supplied procedures don't work, and I have to back to the "old ways." Note what I am complaining about here: someone in an interview would incorrectly reject someone like me, because I can't immediately answer a question which I would if I spent several hours a day coding PL/SQL. Not something I particularly want to do. Shoot, in the big scheme of things, it makes about as much sense as asking obscure questions about INP files. Which leads me to ask: Complete this sequence: Mainframe, minicomputer, micro, network, hyperlink, ??? I maintain that the next major paradigm shift will be... (drum roll please, cue Ziggy Stardust music) writing systems that will last for a long time. Wish I had some cute little word I could put there! But if that is true, writing O stored procedures in PL/SQL is a really bad thing. The usual upgrade path is a really bad thing. Training and certs as now practiced is a bad thing.

Now, let me tell a story. I worked with an excellent DBA, the fellow was supplied by Oracle, had spent years at the site working on their DW. By
all measures he was a great DBA. I come in, and start working on my "little"
part of the system (online aircraft parts inventory and projects, if you care). I see where he had tried to set up a db and failed. Why was this? I can tell you, it wasn't him - others before and after him were also befuddled by it. The software was was out in left field as supplied by the vendor. I had no trouble with it, because I had no illusions it would be "correct" in the Oracle sense, in fact, I was highly entertained by how ridiculous it was written - Assembler on top of Cobol on top of an abstraction layer for db independence, (or was it the other way 'round?) with a bunch of apparently student-written unix and O-package stuff thrown in at various times. I guess, being entertained by that is true geekdom.

>
> > > Your question is analogous to the following:
> > >
> > > Every job listing for Chief of Thoracic Surgery is asking for 3 to 5 years of
> > > experience performing surgery. How are you supposed to get into the business?
> > > Or
> > > Every job listing for Supreme Court Justice is asking for previous experience
> > > as an attorney and judge. How are you supposed to get into the business?
> > >
> >
> > Have you ever seen a job listing for Supreme Court Justice?
>
> Have you ever seen a job listing for DBA without experience?
>
> > > In short ... do your apprenticeship. Pay your dues. There are a lot of good
> > > Oracle developers who are far more qualified than you and they get first shot
> > > at every opportunity which is as it should be.
> >
> > The "pay your dues" stuff is reasonable. However, the part about
> > developers getting first shot at every opportunity is very
> > questionable.
>
> Not in any shop where I've ever worked. Not not with respect to any hiring decision I ever make.

Well, that just shows the limitations of your experience, doesn't it?

>
> I just blew someone out of an interview that had three previous years of DBA experience on their resume
> because they couldn't (and I swear it is true) write an inner join and turn it into an outer join and
> explain the impact.

OK, that's bad.

>
> >
> > In many cases it is as others have posted, management has decided to
> > move someone into the empty spot, and often that someone is "excessed"
> > from another group, or some PC guy. When I was vendor-supporting
> > DBA's, just about every one was new (probably some skew there because
> > experienced ones wouldn't be asking... but I didn't get the impression
> > that there were many customers like that). When I worked in a big gov
> > shop, it was amazingly stupid. That's the real world.
> >
> > _Should_ developers get first shot? Some DBA's might argue about
> > removing the "first" :-) If the place is big enough for more than one
> > DBA, I would argue only one of them should have come from the recent
> > development ranks, the skills required are much more global, and there
> > should be a spread of experience among the DBA's (both length of time
> > and breadth of experience).
> >
> > And I don't know if my view is again skewed by smaller shops, but to
> > me it appears that most places don't want to have a full-time DBA, as
> > it is considered simply overhead. So you wind up wearing multiple
> > hats. Could be a bit of a pain on your resume, when you spend the
> > majority of your time doing DBA tasks, but are not a "DBA." But if
> > you really love it, it's not like work at all.
> >
> > jg
> > --
> > @home is bogus. "The No. 1 problem today is well-known; system
> > administrators not patching systems with the latest (software)
> > updates." - Joe Mambretti, director of Northwestern University's
> > International Center for Advanced Internet Research
>
> Most smaller shops can't afford a fully qualified DBA. But if they could ... they'd be standing in line
> for them.
>
> It is a question of money ... nothing else.

Money - and bad management. I've seen places that can well afford it, but don't. Usually because they incorrectly consider all workers interchangeable resources.

jg

--
@home is bogus.

Hey, ranting is fun!