Re: Deadly sins againts database performance/scalability

Comments in-line.

Paul Drake wrote:

> you want an account that has some sort of dba role that can alter
> database properties, and have access to the init.ora or spfile.
>
> your existing dba is unavailable, overworked, not competent,
> uninterested or just plain has no clue and nothing to offer in terms
> of background, experience, etc.
>
> I can see you wanting to be able to make changes to the database
> config so as to help optimize how the app code runs. you want the
> project to succeed.
>
> I could also see where an existing dba (group) might not want you
> tweaking parameters without justification, this sounds not so much
> like a communal development environment, but a play area for you to
> hone your dba skills. personal workstation and laptops are a great
> place to experiment. If you have to support a horde of developers with
> this dev db, it might not be the best place for you to develop/hone
> your dba skills.

If I may jump in for Galen here, he's probably smarter than me and in bed ... what you've written is not the point at all. But your attitude is exactly what causes developers so much grief.

Developers don't have machines at home that are like those at work. I don't know a single developer who has a personal HP9000 (well I have two but that's not the point a do mostly DBA work and teaching). Nor are there many with Sun 450s or machines running AIX or OS/390, etc. So lets not pretend one can learn on their personal PC that which will best optimize an application in a business environment.

> one possible approach is to get say 10% of your time allocated to the
> dba group, and gain their respect, work with them to optimize your
> development environment.

Another possible approach would be for one of you that believes in locking down a development environment to put forth actual, factual examples of things that have gone wrong. And I don't mean things that potentially could or might go wrong. And I don't mean urban myths you heard about at a Christmas party. I mean things that actually have happened to you.

Much like the detectives on Dragnet ... I want the facts. ;-)

> another possible approach is to completely break from having any
> involvement from your dbas employed by your company. you will likely
> lose out on assistance that they could provide.

Funny cooperation isn't one of your possible approaches. Why? My suggested approach, and the one I use is to ask for justification and grant anything and everything that will not cause a problem when the code movest to test. If they mess it up ... that is their right. Interestingly though ... it has been more than three years since I've seen a development box blow up because of what developers did. And the last one that did was when a junior member of the DBA team tried to show off at a level above her abilities.

> most likely, you do not want a development crew making database
> changes en masse.

And if they have management why would they?

> if anything, give them alter session so that they can muck with things
> as far as their session is concerned, but not muck with other user's
> sessions, or instance-wide settings.

Why? Provide one actual example of something that actually happened and how this caused a problem.

> I am honestly trying to be constructive here, and might have as much
> to gain from these thought experiments as you.
>
> Pd

I'm sure you are. But the divide between you and me (and I am mostly a DBA) is as wide as the Grand Canyon. Between you and most deveopers the size of the Pacific Ocean.

I understand your concern about people doing stupid things. I have the same concerns. But I am still asking for and awaiting actual examples of things that have gone wrong. From where I sit you have crossed the line from concern into paranoia.

And not once has anyone addressed my statements about granting access to developers to the v$ magic views and many of the DBMS and UTL packages and I would bet a Latte' at Starbucks you don't give your developers access to them either.

-- 
Daniel Morgan
http://www.outreach.washington.edu/ext/certificates/oad/oad_crs.asp
http://www.outreach.washington.edu/ext/certificates/aoa/aoa_crs.asp
damorgan_at_x.washington.edu
(replace 'x' with a 'u' to reply)