Re: Tuning: which applications are writing too much

Jonathan Lewis wrote:
"Gerard H. Pille" <ghp@xxxxxxxxx> wrote in message news:44098a6f$0$31480$ba620e4c@xxxxxxxxxxxxxxxxx

Hallo,

our applications - or some of them - are generating more data than our hardware can stomach. The question is of course which ones. Should be easy to answer if I can determine which tables are written to most.

Don't tell me to RTFM, I know all about the F, and I'm starting with the Oracle manuals the minute I've posted this.

Any hints are well come, even the obvious ones, I can always have overlooked something stupid, haven't been doing any general tuning for quite a while now.

Thanks,

Gerard

PS: Most of the databases are 9.2.0.7, all are version 9.



Since you are on 9i, you can make use of v$segstat.
For each active segment it holds a dozen or so stats,
the names are listed in v$segstat_name. These inclue:

db block changes
physical writes
physical writes direct

Be a little cautious with v$segstat, using it used to cause
a memory leak in the SGA, but an "order by" clause seemed
to stop the leak. Select where value greater than some sensible
limit for the statistic you want.

Don't use v$segment_statistics - which is the expensive version
of the same thing.



Thanks Jonathan, I believe that is what I'm looking for.

Gerard
.

Relevant Pages

  • Tuning: which applications are writing too much
    ... our applications - or some of them - are generating more data than our hardware can stomach. ... Most of the databases are 9.2.0.7, ...
    (comp.databases.oracle.server)
  • Re: Tuning: which applications are writing too much
    ... hardware can stomach. ... Oracle manuals the minute I've posted this. ... a memory leak in the SGA, but an "order by" clause seemed ...
    (comp.databases.oracle.server)
  • Re: Small Business Server 2003 lockups
    ... This sounds to me like some memory leak. ... It is a full system lockup as we cannot ... veterinary database) and Quicken. ... scans and hardware diagnostics. ...
    (microsoft.public.windows.server.sbs)
  • Re: malloc trouble
    ... Hmm...how would I do the realloc later then if the space was never ... > lead to an immediate exit, as per the code below), you should assign the ... > memory leak as you've lost the pointer to the original buffer. ... that you mean memory that was never freed for other applications to ...
    (comp.lang.c)
  • Memory Allocation/Accounting bug?
    ... I was doing some leaktests with our applications and couldn't really ... figure out any memory leak. ... PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ...
    (Linux-Kernel)