RE: DataGuard - Maximum Availability mode vs sync detection
- From: "Ric Van Dyke" <ric.van.dyke@xxxxxxxxxx>
- To: <Laimutis.Nedzinskas@xxxxxxxxxxxxx>, <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 19 Jan 2007 12:56:46 -0600
You would have to actively be writing out some sort of log from the
primary and standby to some yet other location to be sure. Since lose
of either site might cause you to lose the log. All the data is there to
figure it out, but if one system is totally gone, then you will only
have half the picture.
And you are absolutely correct. If you are in MAX Availability and the
link is lost from the Primary to the standby and then you have a
catastrophic loss of the primary (as in the entire system is totally
lost to a fire for example), you will lose some data. And you wouldn't
really know how much.
If the system is absolutely mission critical you should have more then
one standby (three is a good number). And each standby in a different
time zone from the primary and each other would be best, each with its
own distinct connection to the primary. Is that expensive? You bet, but
how much will it cost if you lose the system? If the business can live
with some lose of data, then no need to go to this extent.
Ric Van Dyke
Hotsos Enterprises
-----------------------
Hotsos Symposium March 4-8, 2007. Be there.
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Laimutis Nedzinskas
Sent: Friday, January 19, 2007 6:05 AM
To: oracle-l@xxxxxxxxxxxxx
Subject: DataGuard - Maximum Availability mode vs sync detection
Is it possible to know if standby is synchronized with primary when DG
is configured in Maximum Availability mode?
As I understand - it is not possible.
If primary dies for sure(absolutely no access to any primary files) then
there is no way of knowing if it was operating in Maximum
Availability(meaning standby is in sync) or downgraded into Maximum
Performance mode(meaning standby is not in sync.)
What are possible solutions to this problem?
I see just one: some data of the primary db must survive the primary
crash.
For example, the fact that primary went into downgraded Maximum
Performance mode is recorded into background alert log. If we have
access to this log then we can find in what mode the primary database
operated just before the death. Alternatively, control file is the most
credible source.
The question is: what is the best way to preserve this critical primary
db data so that it is readily available for the third party which
decides the failover? I can think of some shared storage for both
control files and background logs.
Fyrirvari/Disclaimer
http://www.landsbanki.is/disclaimer
--
//www.freelists.org/webpage/oracle-l
--
//www.freelists.org/webpage/oracle-l
Other related posts:
