Re: Archive process too slow - please help

From: A. Paul Singer <singerap_at_powergrid.electriciti.com>
Date: 6 Jan 1995 03:28:34 GMT
Message-ID: <3eidd2$buc_at_arc.electriciti.com>


howerter_at_world.std.com (Bradley W Howerter) wrote:

>

> My archive background process cannot keep up with the database writer.
> Even though we have more than 1Gbyte of redologs, I end up getting
> the error "All online redologs needed archiving" because the
> archive process is too slow, and it is still working on the
> redolog file that is needed by the database writer.
>
> We figure we could work around this by adding even more disk space
> for the redologs, but I'm wondering if there are any other alternatives.
>
> This is Oracle V7.1.3 on Sequent PTX.

One gig of redo seems quit excesive. You must be pumping a very large amount of data through your database.

I have a similar envirnment. My main production instance has 6 10meg online redo logs ( on avarge we only cycle one log per day) These are archive to disk and then moved to tape by a seperate job.

You need to seperate the destination of the archived redo logs from that of the on line redo logs. They should be on seperate disks and or SVM volumes on seperate disks and or seperate QCIC I/O channels.

You do not want to place any of these file systems in STRIPED SVM partitions because striped volums do not perform as well for sequential writes. ( I have this from Sequent and I have demonstraited it imperically with redo logs. This does not seem to hold true for raw read from striped partitions.)

Having your redo logs in raw disk partitions is also a performance benifite. (between 5 & 20 percent improvement depending on things wiser men than I could tell you )

One thing to be aware of. I have notice that if the destination for the archive becomes full than the archiver really hammers the system.

The result is that reading the archived redo logs out of the destination volume is very slow. Normally I can sustaing about 300k/sec to my tape (this is the speed of the tape) even if the archiver is writing a new redo log to the archive disk volume.

If the archive volume is full and the archiver is trying to write another I can only sustain 90k/sec. reading out of the volume.

The moral is, if you archive to disk make shure you never let your archive destination volume get full.

Good luck. You can also contact me at singerap_at_songs.sce.com. Received on Fri Jan 06 1995 - 04:28:34 CET

Original text of this message