Re: Boss is asking for 'root cause analysis'

On Thu, 01 May 2003 05:56:58 GMT, "Kenneth A Kauffman" <kkauffman_at_nospam.headfog.com> wrote:

>
>"Ed Stevens" <nospam_at_noway.nohow> wrote in message
>news:rb70bv0n0be925sedeijutjq78lmvh72l6_at_4ax.com...
>> On Wed, 30 Apr 2003 18:37:56 GMT, "Kenneth A Kauffman"
>> <kkauffman_at_nospam.headfog.com> wrote:
>>
>> >
>> >"Ed Stevens" <nospam_at_noway.nohow> wrote in message
>> >news:bbsvavoe46gl1rkd0q0pps8kqkadbm5l83_at_4ax.com...
>> >> On Wed, 30 Apr 2003 15:29:20 +0100, "Niall Litchfield"
>> >> <n-litchfield_at_audit-commission.gov.uk> wrote:
>> >>
>> >> >Unless you have some sort of performance monitoring data, I suspect
>that
>> >> >finding the root cause will be impossible.
>> >> >
>> >> >If you have , say, statspack or ultbstat/utlestat reports covering the
>> >> >relevant period, or explain plans for the problem queries before and
>> >after
>> >> >it will be pretty difficult to see what was going on.
>> >> >
>> >> >I'd strongly suspect the dataload causing the access plans to become
>> >> >inappropriate but without seeing the access plans it is pretty well
>> >> >impossible to tell.
>> >> >
>> >> >The other question to ask is wether the reporting tools/queries had
>been
>> >> >changed/upgraded in anyway.
>> >>
>> >>
>> >> Niall,
>> >>
>> >> Thanks for the reply. You add weight to what I already suspected.
>> >> From my end, I assume that I don't know what it is that I don't know,
>> >> so am always looking for confirmation or refutation. This whole app
>> >> has been a severly underfunded disaster waiting to happen.
>> >>
>> >
>> >ISBN: 007222360X
>> >Oracle9i High-Performance Tuning with STATSPACK
>> >
>> >Fairly decent book and also explains how to automate capturing statistics
>on
>> >a scheduled basis, including system resources.
>> >
>> >ken k
>> >
>>
>> Oracle 8.0 -- statspack not available until 8.1. (Well, I understand
>> there is a kludge backport of statspack). Utlbstat/estat *is*
>> available, but doesn't help much *after* the fact.
>>
>> It just occurred to me that I *could* at least run an explain both
>> with and without the STAR_TRANSFORMATION_ENABLED parm. That would
>> probably shed *some* light on *one* aspect of the problem. But I know
>> that when the mgrs start asking for root cause analysis, they want to
>> identify the *exact* problem and a firm preventative measure.
>
>Apologies sir. I missed the version and was thinking 8i the whole time.
>
>ken k
>

No apologies needed. Received on Thu May 01 2003 - 10:15:37 CDT