Re: Dataguard: Max Availability vs. Max Performance
- From: Carel-Jan Engel <cjpengel.dbalert@xxxxxxxxx>
- To: oracle-l@xxxxxxxxxxxxx
- Date: Thu, 15 Apr 2004 00:55:42 +0200
Hi Walt,
It depends.
What performance are you talking about? Apologies for not understanding
your 'page truns' phrase. What's 25. million turns? hits or changes?
changes or transactions? Even then, 2,500,000 transactions/month is an
average of 0.96 transaction /second. It will not be of significant
influence on your queries.
The most important part is your network latency, and then the bandwith.
I've several customers running with DG configurations in LGWR/SYNC mode.
One of them with the standby in another building, appr. 0.75 miles away. No
problem. They have GB fiber connections, but everything runs fine. Others
have servers sitting in two neighbouring 19" racks, or separated but in the
same building. No performance pains. I must admit, because the systems run
fine, and everyone is happy, we haven't tested the actual limit. But, we
simply activated MaxAv. because it worked, although a LGWR/ASYNC setup
would be enough. Small data loss is allowed. So, whenever the limits are
hit, we will fall back on ASYNC redo log forwarding.
The only way to find out the limit for your particular system is testing. I
prefer setting up a neat, small database in a DG environment, just to have
it available to play around with it. This contributes significantly to the
confidence of the local DBA. And, when you like, you can use it to push the
system to the limits.
In a LGWR/SYNC setup I allways advise to install separated NIC's and, if
possible, an isolated segment to separate DG-network traffic from
application traffic.
There is a paper on metalink, doc.id. 240874.1, giving some metrics. I
wouldn't rely on the metrics, but it gives an idea how to set up the test
for your environment.
Regards, Carel-Jan
===
If you think education is expensive, try ignorance. (Derek Bok)
===
At 06:28 PM 4/14/2004, you wrote:
>I'm setting up a Dataguard configuration (Dell 6650's, RedHat AS 3.0,
>9.2.0.4) and was curious about the statement in the Dataguard Concepts
>and Administration manual stating that Max Availibility "presents a
>potential response time degradation".
>I am aware of why there might be a performance degradation, but I was
>wondering if anyone was using Dataguard with Max Availability. If so,
>have you seen any kind of degradation?
>
>Are there any metrics around showing what the potential degradation
>might be? The primary machine in our Dataguard configuration will be the
>back-end of a hosted web site and will be serving up about 2.5 million
>page turns a month.
>
>Is there any way I can determine what sort of performance hit will occur
>if I set Dataguard to Max Availability?
>
>The customers on this machine are really anal about performance for some
>reason and I don't want to make'em mad.
>
>Thanks,
>
>--Walt Weaver
>
> Bozeman, Montana
>
>
>----------------------------------------------------------------
>Please see the official ORACLE-L FAQ: http://www.orafaq.com
>----------------------------------------------------------------
>To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
>put 'unsubscribe' in the subject line.
>--
>Archives are at //www.freelists.org/archives/oracle-l/
>FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
>-----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
