Re: Frequent log switches during sqlldr batch jobs

From: Thorsten Jens <thojens_at_gmx.de>
Date: 3 Feb 2005 01:27:28 -0800
Message-ID: <1107422848.137137.95370@c13g2000cwb.googlegroups.com>

Frank van Bortel wrote:

> Ah, no: 5 hourly during "normal" loads. And it's only an indication;
> I wouldn't start to worry if it's 20 in your case - all runs well
> with that aspect (no timeouts, as you stated).
>
> As said, there's some risk/time to recover (MTTR - mean time to
> recover) involved here, too.
>
> As you still haven't revealed your peak load, nor see any timeouts,
> I would not change a thing, given the information available

Here's an example of our workload (log switches/hour):

DAY 00 01 02 03 04 05 06 07 08 ----- ---- ---- ---- ---- ---- ---- ---- ---- ----

01/31    3    0    0    0    0    0    0    1   12
02/01    3    0  101   24   65    0    0    0    2
02/02    3    0  100   21   67    1    0    6    3
02/03    3    0   96    6   70   18    0    1    0

09 10 11 12 13 14 15 16
---- ---- ---- ---- ---- ---- ---- ----
1 1 5 7 2 4 1 1
1 0 1 25 9 6 5 0
2 7 18 10 15 5 3 2
0 0 0 0 0 0 0 0

17 18 19 20 21 22 23 Total ---- ---- ---- ---- ---- ---- ---- --------

1    1    3    0    1    2    2       48
2    1    3    0    0    3    1      252
1    1    4    0    0    2    1      272
0    0    0    0    0    0    0      194

>From the responses I guess there's no urgent need to worry.

Thorsten Received on Thu Feb 03 2005 - 03:27:28 CST