Re: Are Oracle DBAs trivialized?

From: DA Morgan <damorgan_at_x.washington.edu>
Date: Tue, 19 Apr 2005 11:41:45 -0700
Message-ID: <1113935873.558020@yasure>

Jim Smith wrote:

>> In my experience likely *every hand made* backup script is less robust
>> than RMAN.

>
> Not everyone who isn't using RMAN is using hand-made scripts.

But those that aren't are paying a lot of money to avoid using that which their company already purchased.

> There are
> other solutions, but in any case a carefully written script by an
> experienced DBA can certainly be robust. DBAs were writing backup
> scripts for a long time before RMAN was introduced and probably got
> quite good at it.

'Probably' is an excellent choice of words. I have yet to see the DBA whose script was put into change management and given to an independent team for testing and certification. And those that do work ... work only so long as the DBA has tribal knowledge of what is where and manually changes it each time a change is made to the database.

Want to know how many times I have seen these carefully crafted scripts fail when a datafile was moved or added?

>> So when you are talking about *critical* components,
>> chances are you might already have one.

>
> You misunderstand the meaning of "critical". It means that a component
> is (very) important, not that it is more likely to fail.

Your definition but not one generally in use. How about critical as in a show-stopper if it doesn't work perfectly. It may or may not be likely to fail. But when it does, such as an inability to restore, the system stays down.

>> You aren't aware of that
>> because self-developed solutions are often justed *trusted* instead of
>> *tested*,

>
> Someone who hasn't tested their current solution is also unlikely to
> test their rman solution.

Agreed. Those that don't test deserve their own special place in hades.

>> and it often is quite easy to break them.

>

>> One little example
>> is a script where the list of files to be backuped had to be
>> maintained in a separate configuration file. As you claim you know
>> everything so much better, you could probably, as a quiz question,
>> identify why this is a crap solution, and identify the costs involved.
>>

> It is probably equally easy to develop a solution using rman which is
> fatally flawed.

But there are things I can do in RMAN you can not do with a script. And lets start by discussing block recovery.

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace 'x' with 'u' to respond)

Received on Tue Apr 19 2005 - 13:41:45 CDT