Re: too many logswitch ?

In article <942440807.28148.1.nnrp-14.9e984b29_at_news.demon.co.uk>,   "Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote:
>
> I would have expected 1,500 to 2,000 bytes
> per row of redo if you have been committing
> once every row, so you figure seems to be
> about 5 times too much.
>
> Are you sure nothing else was going on,
> and that there were no indexes on the table ?
>
> Are you sure that the redo block size is
> only 512 bytes on your platform ?
>
> What do the 'redo size' and 'redo wastage'

jonathan, thanks for the reply.here is the redo info. redo size is around 1g and redo wastage is around 28m. and this number is reasonable.
Is there any other problematic parameter in the list - i.e. redo small copy ?
thanks again
ted chyn

STATISTIC# NAME

     CLASS VALUE
---------- ----------

       104 redo blocks written
         2    2050012

        99 redo buffer allocation retries
         2       1944

        96 redo entries
         2    3869928

       106 redo log space requests
         2        670

       109 redo ordering marks
         2       2802

        97 redo size
         2  988026344

       100 redo small copies
         2    3871272

        47 redo synch writes
         8    1248147

       101 redo wastage
         2   28352548

       103 redo writes
         2     101004

> figures look like ? And how many redo blocks
> were written ? (see v$sysstat)
>
> --
>
> Jonathan Lewis
> Yet another Oracle-related web site: http://www.jlcomp.demon.co.uk
>
> tedchyn_at_yahoo.com wrote in message <80hoct$574$1_at_nnrp1.deja.com>...
> >Sir,
> >
> >1. Inserting into a table(max rowlength 300 bytes,NO index) via odbc
> >driver.
> >2. after insertion table size is 7m with 100,000 rows
> >
> >3. logswitch 170 times
> > redo log size 5m
> > total logfile consumption 750m(170 x 5)
> > log_checkpoint_timeout is set to 0
> > log_checkpoint_interval is set to 100,000 os block size(512bytes)
> >
> >4. the ratio of redo cosumption to size of table is almost 100.
> >
> >questions,
> >1. what control log switch ? if logswitch takes place when it is
filling
> >up then one would assume during switch redo is almost full?
> >2. why redo conumption is so high for the number bytes inserted into
> > the table ?
> >
>
>

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Mon Nov 15 1999 - 12:09:27 CST