Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Why are people so afraid of underscore parameters ?

Re: Why are people so afraid of underscore parameters ?

From: Richard Foote <richard.foote_at_bigpond.com>
Date: Fri, 23 Aug 2002 10:29:57 +1000
Message-ID: <c6f99.12411$g9.40345@newsfeeds.bigpond.com>


Hi Bass,

Aren't some people weird.

I'm personally afraid of the dark, spiders, any insect larger than my little finger, heights, death, Mike Tyson (closely related to the previous one I know), sharks that I don't see, bagpipes, Pete Sharman and people in general who look at me kinda funny.

But to be afraid of _parameters, well that's really quite bizarre ;)

Actually, I think both your Boss, yourself and Oracle are all correct in your own ways.

I would suggest, that the vast majority of the _parameters in the vast majority of cases do not have to be touched at all. I would also suggest that the vast majority of the parameters if modified would have a more probable chance of doing harm than good. If most of these were to be documented, I can tell you now, the number of calls and tars that Oracle Support would have to deal with would escalate to scary proportions. So I completely agree with Oracle 'hiding' these parameters and documenting and hence encouraging us to use and modify only those parameters that have a greater chance of making a positive outcome in our environments.

Therefore I agree with the cautious approach being adopted by your boss. If you never touch an _parameter, you're going to reduce the chances of any significant stuff-ups, stuff-ups that Oracle Support may not take too kindly on.

However, the are some situations when the modification of these parameters is beneficial and in fact recommend. There was a thread here not too long ago on the positive impact of changing the _spin_count parameter. Usually shouldn't be touched, probably would hurt rather than improve performance if touched without due care but could result in a positive return if modified in certain situations appropriately. Therefore, you are potentially correct to suggest the odd tweak here or there and your boss is being somewhat pedantic by saying never. However, such changes should have the following 'steps' associated with them.

  1. A clear understanding on the use of the particular parameter and it's possible effects.
  2. A clear objective for the changing of such a parameter
  3. A demonstrated benefit in (preferably in a test environment first) for such a change and a clear demonstration of no significant side-effects
  4. A 'chat' to Oracle Support regarding the change and some form of confirmation that your tack has some merit and chance of success
  5. Clear documentation on the parameter change, date of effect, reasoning for change and verifiable effect of the change

Although all the above are not strictly required and in some cases not practical or achievable, at least going through the thought process suggests you haven't made a change for change sake and should satisfy your boss on your motives and course of action.

My thoughts

Cheers

Richard

"Bass Chorng" <bchorng_at_yahoo.com> wrote in message news:bd9a9a76.0208221016.3f3f9dcf_at_posting.google.com...
> My boss does not allow DBAs to use any underscore parameters.
> He seems to be unreasonably freaked out upon hearing one.
>
> Some also advise that you should never use it without being
> instructed by Oracle Support.
>
> A lot of good tuning parameters in 8i have gone undergound in 9i.
> Some parameters have stayed undergound since Oracle 7 but really
> should be normalized, such as _trace_files_public which I see
> no reason to be hidden. I have never seen a single
> production site not using that (except current one). It is a no
> brainer parameter.
>
> I know hidden parameters were initially used by developers.
> They got released without adequate documentation. Later on
> Oracle moves normal parameters undergound in new releases
> because the functionality of such parameters were deminished or
> probably just to reduce the number of visible parameters for
> better looks and easier documentation.
>
> In 8.0.5, there are 141 underscore parameters, in 8.1.7, it
> becomes 301, in 9.2.0, it is 544. All it tells me is, that
> Oracle just keeps on moving parameters undergound to keep
> the other side slim.
>
> But to say that Oracle does NOT support them is just not true.
> ( This has nothing to do with some specific underscore parameters
> that they do not support such as _disable_logging )
> I have used numours underscore parameters per Oracle's suggestions.
>
> I think if you know what a parameter does and you know that is
> what you need, there really is no need to treat tham as taboos.
> You can do damage to your database too with normal parameters if you
> don't know what you are doing. Oracle obsoletes parameters on either
> side the same way.
>
> If you disagree, I like to hear your comments.
Received on Thu Aug 22 2002 - 19:29:57 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US