Re: Are Oracle DBAs trivialized?

In message <c7e08a19.0504101902.6ff029dd_at_posting.google.com>, Domenic <domenicg_at_hotmail.com> writes
>I'd like everyone's thoughts on this ...
>
>More and more I am finding that the role of the Oracle DBA is not
>taken as seriously as it once was.
>
>Having worked for many software companies in the last few years I see
>the same pattern over and over -- most shops refuse to stay current
>with Oracle and patchsets -- they're still running 8i, don't want to
>use Oracle-specific features (clusters, IOTs, etc.) for fear it will
>lock them into Oracle, and let the developers model and write their
>own DDL.
>

There are two different issues here - development methods/styles and use of Oracle features, but the both ultimately come down to cost/benefit.

The knack is not to argue in technical terms, but to point out the benefits and costs. Admittedly this is not always straightforward.

>Most of the time I feel they only want a DBA to put out fires.
>
>Here are some recent examples ...
>
>- I had to pull teeth to get my current company to switch to RMAN.

Why should they switch to RMAN? The currently backups presumably work fine, and there will be a significant cost in terms of DBA (and other time) in making the switch and testing it.

Do the benefits of RMAN outweigh these costs (and the risks associated with the disruption of something as fundamental as backups?) Almost certainly, but some figures in terms of DBA time saved, shorter backup windows, faster restores etc, would help the argument.
>
>- I got laughed at for naming NOT NULL constraints in my CREATE TABLE
>scripts
>

I'm afraid I would laugh too. Life's too short.

>- I had a one of my designs tossed out because it had triggers,
>clusters, and deferrable FK constraints in it.
>

Why? Non-portability? Performance concerns? Not making your developers use these things is one thing, actively rejecting them is another.

>- I have been telling management for almost a year to get off 8i --
>falls on deaf ears of course
>

This is tricky. There are huge costs to upgrading, but the risks of hitting a bug for which there is no fix are unquantifiable. In terms of the benefits of later versions, again they need to be quantified. If the shop is just using the database as a persistence store, then benefits other than performance improvements are hard to see.

>- I get apps tossed at me with hundreds of poorly written queries
>(correlated subqueries everywhere, NOT EXISTS, etc.) that I have to
>rewrite for speed.
>

This ought just to be part of the performance testing of the app anyway, whether you do the re-writing or the developers. (I accept that performance testing probably isn't done but one can dream.)

>- I get crap from the UNIX SAs at just the thought using 10g ASM when
>we finally upgrade, by that time Oracle 13 will be out.
>

This is just an old fashioned turf war. (See the thread (Are UNIX SAs trivialised?)[1]. But again, there are probably benefits in terms of their resources which could be pointed out.

>- I have developers who commit every SQL statement and throw highly
>denormalized tables at me, but management still feels they should NOT
>have to come to me to get a table designed.
>

>- HA requirements but they refuse to use RAC or Data Guard.
>

>This is not just my current company, but they all seem to be the same
>lately. It seems that the Java shops operate this way more than
>anyone else -- very hostile attitude towards Oracle and DBAs in
>general.
>
>Is it just me, or does this go on everywhere?
>
>Domenic.

-- 
Jim Smith
Because of their persistent net abuse, I ignore mail from
these domains (among others) .yahoo.com .hotmail.com .kr .cn .tw
For an explanation see <http://www.jimsmith.demon.co.uk/spam>