Re: Monitoring question

Have a look at V$SYSTEM_EVENT, V$SESSION_EVENT, and V$SESSION_WAIT to see what is causing your bottlenecks. V$SESSION_WAIT can tell you *exactly* what each session is waiting on at that exact time. V$SYSTEM_EVENT will tell you what every process has waited on since instance startup. V$SESSION_EVENT will tell you what a specific session has waited on since the session started.

This gives you much more detailed information than the GUI tools will tell you.

HTH,
Brian

Jan Gelbrich wrote:
>
> Hi,
>
> I hope my question is not *too* silly,
> but I would like to know if there is any view inside the data dictionary
> to see in real time which user is producing temporary bottlenecks in the
> database -
>
> the point is:
> I want to know what is happening *at that very time*
> when I see jams and hear users start complaining,
> but most DD Views like v$sysstat show _accumulated data_ ...
> I went through many v$* views in the manuals,
> but I was not successful to find something leading to this direction.
> and v$sqlarea show the SQL, but not the loads produced by it;
> I see just placeholders for bind variables.
>
> A fellow in my team just wondered
> why is there no similar functionality
> compared to the System Monitor on Windows NT or such,
> or top on linux, to see what is happening in time.
> You cannot even see any progress information in SQL+ ...
> ... for the moment I answered him well Oracle is just not M$
> ... but in a way he is right and he made me curious. It should be
> very important to any DBA to see session loads at one hit, isnīt it ?
> And I began to wonder how DBAs with much bigger dbs than mine,
> and thousands of user sessions are doing this ...
>
> Sometimes, mostly on Monday and Friday mornings,
> I have a jam situation for about 1 hour
> where everybody is waiting for app responding,
> but I have no blockings anymore. It is just 30+ users active at the same
> time
> using a couple of inhouse apps (Forms).
> So I cannot see who is to "blame" *in that very moment*
> (of course not the user, but the app (s)he is using ...).
>
> I know there are many possible reasons:
> data model (20%), app (50%), network (10%), server parameters (10%), old
> hardware etc.
> and I want to isolate the problem,
> but yet I am still lacking some techniques to do so.
> I want to overcome guessing it might be this or that app ...
> so v$session and v$sqlarea etc, is not enough for analysis.
>
> I admit that I am still not very experienced in monitoring,
> so I appreciate any comments.
>
> Is there a startegy or some points to start from in complex situations like
> this ?
> Thank You in advance.
>
> Jan
>
> My system:
> Oracle 8.1.7 EE on a central AIX-Server 4.3.3, dedicated server, C/S network
> to 20 local subsediaries.
> Oracle Forms & Reports inhouse app. Bind variables are used all over the
> place,
> and a fairly small DB of 20 GB, 200 users, jam starting when 30+ active.

-- 
===================================================================

Brian Peasland
oracle_dba_at_remove_spam.peasland.com

Remove the "remove_spam." from the email address to email me.


"I can give it to you cheap, quick, and good. Now pick two out of
 the three"