With all due respect, I disagree with the assertion that aggregate statistics are useless. While aggregated statistics have problems in pinpointing specific problems (which isn't their intent anyway), they can point the way toward problems in a general fashion, while serving admirably to debunk groundless speculation as well. The fact is that aggregated information and specific information provide a continuity to be used in tandem, and they should each be understood for their strengths and weaknesses. Take political campaigns as an analogy. While media polls do not provide information about why people vote, they most certainly do accurately predict how the vote will tally. If you want to know why, you talk to individuals. If you want to know the trend, you poll masses. Trends cannot be detected from individual interviews any easier or more accurately than specific causes can be detected from mass polling. With the volume of misinformation about the causes of performance problems on Oracle-based systems, it is helpful to have information at hand going back over a period of time that points towards certain problems and conclusively eliminates others. Keeping management off one's back should not be underrated. IT management is eminently capable of wasting vast amounts of time and effort due to something read in a magazine or heard from a friend "in the know". Countering the gibberish with fast facts is helpful in keeping an organization focused on what is vital. Personally, I spend a fair amount of time applying data warehousing techniques to the data collected by STATSPACK, in addition to reading trace files. I use both tools, each where applicable. If I had to choose one tool or the other exclusively, then I agree that tracing is much more accurate and can feasibly be adapted to trend analysis if necessary. But I don't have to choose, so I prefer to use the best tool for the task at hand. In statistical analysis terms, analyzing trace files is "directed analysis", requiring preparation and prior direction. Using the same terms, analyzing collected aggregated statistics is "undirected analysis", which can be performed before a direction is given or even known. Another broad term for undirected analysis is "data mining", although that is certainly too grandiose a term for STATSPACK and completely unsuitable for the default report provided by STATSPACK. Just my $0.02... -Tim on 2/2/04 7:32 AM, Freeman, Donald at dofreeman@xxxxxxxxxxx wrote: > I'm a relatively new DBA but I have 30 years electronic engineering = > experience. I'm used to tools that work and actually measure what they = > purport to. I got all excited about the capacity planner about 6 months = > ago and asked the same questions you are asking now. Mostly, nobody is = > using it. It becomes a headache itself, the agent fails and causes you = > grief. I don't think you'll find much usefulness in it. When you start = > troubleshooting it won't give you anything helpful.=20 > > After reading Carey Milsaps Optimizing Oracle Performance I am less than = > thrilled with statspack also. You can't solve (or even determine) a = > particular problems origin while looking at aggregate values. > > The main value of these things is to provide a comfort level and = > distraction to management. Attach your statspack report to an email and = > send it to your boss. It should keep him (or her) busy for some time = > while you work on the database. > > > > -----Original Message----- > From: oracle-l-bounce@xxxxxxxxxxxxx > [mailto:oracle-l-bounce@xxxxxxxxxxxxx]On Behalf Of > Luc.Demanche@xxxxxxxxxxxxxxx > Sent: Monday, February 02, 2004 9:19 AM > To: oracledba@xxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx > Subject: Capacity Planner from OEM VS Statspack > > > Hi DBA, > > I'm starting to take a look at the "Capacity Planner" tool from the > Diagnostics Pack. > Great tool, collects info on lots of interesting statistics ... =20 > from databases=20 > - Response time > - Wait events > - I/O > - Storage=20 > - ...=09 > and servers > - CPU > - Memory > - File system > - ... > =20 > Good Report tool, create graphics and you can even do a trend=20 > analysis...=20 > > I have two questions: > 1- Are a lot of you using it? > 2- Does STATSPACK become less usefull? I would keep STATSPACK for the = > SQL > level. Capacity Planner doesn't seem to handler that level. Right? > > Thanks > Luc > > > --------- > Luc Demanche > AstraZeneca R&D Montreal > Oracle Database Administrator > 514.832.3200 x2356 > > ---------------------------------------------------------------- > Please see the official ORACLE-L FAQ: http://www.orafaq.com > ---------------------------------------------------------------- > To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx > put 'unsubscribe' in the subject line. > -- > Archives are at //www.freelists.org/archives/oracle-l/ > FAQ is at //www.freelists.org/help/fom-serve/cache/1.html > ----------------------------------------------------------------- > ---------------------------------------------------------------- > Please see the official ORACLE-L FAQ: http://www.orafaq.com > ---------------------------------------------------------------- > To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx > put 'unsubscribe' in the subject line. > -- > Archives are at //www.freelists.org/archives/oracle-l/ > FAQ is at //www.freelists.org/help/fom-serve/cache/1.html > ----------------------------------------------------------------- ---------------------------------------------------------------- Please see the official ORACLE-L FAQ: http://www.orafaq.com ---------------------------------------------------------------- To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx put 'unsubscribe' in the subject line. -- Archives are at //www.freelists.org/archives/oracle-l/ FAQ is at //www.freelists.org/help/fom-serve/cache/1.html -----------------------------------------------------------------