Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Why? Oracle 7.3 Server stalls Client

Re: Why? Oracle 7.3 Server stalls Client

From: <andreas.prusch_at_sintec.de>
Date: 1998/07/17
Message-ID: <6onj4m$jn3$1@nnrp1.dejanews.com>#1/1

The behaviour is a result of a couple of parameters and of the size of the redo-log files. The client is waiting for checkpoints. This occurs when log-file switching is faster then db-writer's checkpoint activity. The log-writer cannot reuse a log-file which corresponding checkpoint is not completed by db-writer (For more information see manuals).Increase the size of the redo logs and increase groups to at least 4 and eleminate additional checkpoints by setting log_checkpoint_interval to 8000000000. This will allow db writer to do checkpoint until the logfile is filled. Also increase log_buffer to 1 MB. Remember, log buffer will set in bytes. Then insert-rate will be more constant.

Hope this will help.
Andreas Prusch

In article <01bdb14c$e7a62c60$bf13c2cf_at_soldev>,   "Stevyn Dembo" <sdembo_at_soliton.bc.ca> wrote:
> Hello.
> We are running oracle 7.3.3 on Windows NT3.51/P266/64MB memory/2 disks. We
> have written a tiny looping program in Delphi on a W95 client to see how
> many inserts per minute we can make into a single table (2 columns, primary
> key index, no relations, no triggers). The network is TCP/IP. We want to
> achieve the maximum insert rate that is achievable for a single client
> against the db. We have observed the following situation. The driving
> program runs at full speed for about 200 inserts and then stalls. There is
> a bunch of disk activity on server after the stall, and once the server
> disk activity has subsided, the driving program resumes and sends another
> approx. 200 inserts, stalls,server disk activity, on and on! We have
> monitored the network packets and have observed that no packets are being
> sent during the stall. Stalls are approximately .5 - 1 sec.
>
> The BIG QUESTION is: What steps can we take to eliminate or minimize this
> stalling? Currently we can achieve an insert rate of about 4000-4600
> inserts per minute. Without the stalls, we could almost double the rate.
>
> Thanks in advance to anyone who may have suggestions as to how to resolve
> this issue!!
> Regards.
> Stevyn Dembo
> steve_dembo_at_bc.sympatico.ca
>

-----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum Received on Fri Jul 17 1998 - 00:00:00 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US