Re: Case study for interviewing Oracle DBA

I suspect that a junior DBA should be tested more on what they would
NOT do, rather than what they would do.

#1 - in my experience, is never delete a file.  Always rename it.  If
the database breaks you have a chance to name it back and try to
recover.  If the database is working fine 24 hours later then you can
delete it.

#2 -  Monitor statistics over time.  That way you can see trends
developing and plan a strategy for a fix insteading of waiting for
some crucial threashold to be reached and then trying to fix it in
panic mode.

You might has him/her when would you delete the SYSTEM tablespace.  If
they come up with some BS answer try to keep a straight face.

Do they make a destinction between an instance and a database?  A
non-DBA rarely makes that destinction.  You might ask them how they
would make a cold backup of an instance (if you have persistant RAM it
just might be possible).

Can they identify the difference between a logical volumn and a
physical device when it comes to distributing tablespaces?  Does
Oracle access an index and it's table simultaneously or serially?

What's the best RAID for a database --- a decision support database
and an on-line transaction support database?

What's the maximum number of extents that a segment should have?

Does the person know the difference between Oracle myths and Oracle reality?

I've had these case study tests thrown at me and frankly I don't think
much of them.  There's never enough information to do more than make a
very generalized recommendation.  In the real world these questions
are answered by business and financial policies rather than best DBA
practices.

Mostly you'll want to hire someone who will think and then think again
before they put their fingers on the keyboard.
--
http://www.freelists.org/webpage/oracle-l

Other related posts: