Re: RANT: Re: Boss is asking for 'root cause analysis'

On Thu, 01 May 2003 06:16:38 GMT, "Kenneth A Kauffman" <kkauffman_at_nospam.headfog.com> wrote:

><snip>
>>
>>
>> Ah, and therein lies the problem! There *is* no test system, and no
>> available server big enough to build one. DBA and the application
>> developer have been complaining about this from Day One (this app is
>> about six years old) to no avail. What you suggest is essentially
>> what my mgr was also suggesting before we found the solution. He's
>> the new kid on the block -- been here less than a month -- and was
>> appropriately surprised to find we run this system with no failover,
>> no test system, and only a daily server-level disk backup. All
>> because the client isn't willing to fund the necessary hardware.
>
><rant>
>Testing within production and holding accountability for unseen problems is
>an insane model for any management to buy into. Yet is happens time and
>time again. Grant it, sometimes this occurs because some systems
>administrators do not size appropriately and put together plans that only
>meet the immediate production need. The sickest part of your dilemna is
>that when you state to them that you do not have enough statistics, they
>will hold *you* accountable for not monitoring it appropriately by putting
>the tools necessary in place. Then when you need a system to do growth
>analysis and exploratory research, its not there. Given that you are on
>8.0.6 on NT lets me know that this is a miserly organization. 8.0.6 from a
>performance perspective is adequate, but sincerely not the best you could be
>doing. Then to also have the audicity to not fund failover mechanisms and
>still hold you accountable is ludacris. But then again, you'll get them
>coming back that "with appropriate monitoring" you should be able to catch
>issues and repair them with inconsequential downtime. It may still impact
>production and be down for a bit, but in the grand scheme its
>"inconsequential". I'm also guessing that if you're on 8.0.6 on NT, you are
>on some old hardware with slow arrays (if they're using arrays :).
></rant>

As the OP on this subject . . . . you get no argument from me!

>
>Given what star transformation does, I suppose the best thing to do is say "
>The data had grown to a level where enabling this feature allowed us to
>better utilize our bitmap indexes with the data. The explain plans were
>positively changed in a manner that supported the new data volume and
>usage." Or some gooey crap that makes them all smiley happy. If they can't
>provide the proper tools, you can't provide the proper answers.
>
>ken k
>
Received on Thu May 01 2003 - 10:48:14 CDT