Re: Performance stats retention

Thank you Pap.

Yes, I see that's extendable similar to AWR retention. But my concern was if it will have any possible negative impact and it's not mentioned in that book also. And what is the difference between the exadata stats which are visible in the AWR report(which depends on the AWR retention) VS the ones dependent on the "cell metricHistoryDays" retention?

On Tue, Feb 8, 2022 at 11:54 AM Pap <oracle.developer35_at_gmail.com> wrote:

> If you see page no- 245 in the doc below, it does state that you can alter
> the metricHistoryDays using the command " alter cell
> metricHistoryDays='<N>'".
>
>
> https://www.oswoug.org/OSWOUG/Home_files/expert_oracle_exadata_second_edition.pdf
>
> On Fri, Feb 4, 2022 at 1:49 AM Lok P <loknath.73_at_gmail.com> wrote:
>
>> Hello Listers, In many of our customer databases(combination of Version
>> 11.2 and 19C) we have daily and monthend processes runs and that is why we
>> normally set the AWR retention period to >65 days to have visibility for
>> the last two months back activity , so as to look into any performance
>> issues and compare things back in days.It is helping us significantly in
>> triaging production issues. Though these AWR data were consuming storage
>> space on sysaux tablespace but we have not seen any performance issue as
>> such because of this. We have been asked to do the same for cell metrics
>> too by increasing the retention i.e. 'metrichistorydays' parameter.
>> Currently it default 7days. So wanted to understand from experts here ,
>> 1)If this will have any negative impacts? 2) How will these metrics be
>> different from the ones displayed in the AWR report? As because in AWR
>> report we do see the exadata cell flash cache, disk related stats all the
>> way back till we have the set awr retention period i.e.>65days(even we
>> have metrichistorydays set as default 7 days ).
>>
>> Thanks & Regards
>> Lok
>>
>

--
http://www.freelists.org/webpage/oracle-l