Re: why administrator refuse to give permission on PLUSTRACE

From: Galen Boyer <galen_boyer_at_yahoo.com>
Date: 3 Nov 2007 09:29:03 -0500
Message-ID: <u3avn8bm3.fsf@rcn.com>

On Sat, 03 Nov 2007, damorgan_at_psoug.org wrote:
> Galen Boyer wrote:

>> On Fri, 02 Nov 2007, damorgan_at_psoug.org wrote:
>>
>>> The solution is not for the developer to hack around prod hoping to
>>> stumble on the issue.
>> This is a complete misstatement of what is going on.  The developer
>> is armed with something extremely important, which the dba is not,
>> and that is deep knowledge of how the application was written.  The
>> developer is not hacking around and stumbling, he is making very
>> educated guesses at what could be wrong and checking into those
>> pieces of the codebase.

>
> Jonathan Lewis, Cary Millsap, Jeremiah Wilton, and numerous others I
> could name could walk into your shop right this second, knowing
> absolutely nothing about your business including its name, get DBA
> privs on your production database and give you a list of what is and
> is not the issue in a matter of minutes to hours.

That is exactly the point. Those guys are not walking through the door. Having that qualified of DBAs is very very rare.

> Knowing how to fix it may require some tribal knowledge and that I
> think clearly is something developers should do if the issue is code
> related. But you don't fix things in prod ... you identify them in
> prod. Fix them in dev. Test them in test. Deploy the fix in prod.

Okay, then we are agreeing.

> If you want access to prod it is to identify the problem. For that a
> thorough understanding of wait events trumps your employee id every
> time.

A thoroughly qualified DBA is something I'd rather have. I don't want to be on the production system.

-- 
Galen Boyer

Received on Sat Nov 03 2007 - 09:29:03 CDT