Re: rac vs dataguard

  • From: David Ballester <ballester.david@xxxxxxxxx>
  • To: ksmadduri@xxxxxxxxx
  • Date: Thu, 16 Jul 2009 20:19:11 +0200


I'm agree with the points about application scalability, performance, etc...
but what's about D&R? For performance you can always tune your application /
Oracle / OS layer or add more nodes to the RAC but If you loose the primary
site or the whole RAC ( big disaster or big human error ) you will must:

- locate new hardware ( new host ) or make room on existing one.
- Install software  ( OS, Oracle binaries, patchsets... )
- restore the whole database and recover as far as you can ( lost some of
the newest redo logs ? )
- Start to give service.
- If some data was lost, tell your people to redo it again ( if it's
possible )

With the Data Guard you have it done in a few minutes, and near zero data

If you can survive without giving service for all the hours ( days? ) doing
the previous steps, go on adding only new nodes to the RAC

But if service is important to you, you can spend a little more money on the
iron ( is not so expensive this days ) and mount a Data Guard

You can mount a physical standby to protect the whole service and in the
same host, mount a logical standby for reporting purpouses, using both of
them the same binaries.

Define the resources for the physical standby as low as possible ( it only
will be doing recovery ) and allow more resources for the logical one

If a big disaster occurs, you can activate the physical standby ( modifiying
init to allow more resources  ) and point the logical one to adquire data
from the new primary, as part of your contingence procedure.

What's your opinion?



Other related posts: