Re: Is RAC really HA on Linux
- From: Niall Litchfield <niall.litchfield@xxxxxxxxx>
- To: evans036@xxxxxxxxxxx
- Date: Tue, 14 Sep 2004 09:03:29 +0100
Comments in-line
On Mon, 13 Sep 2004 16:14:12 -0400, Stephen Evans <evans036@xxxxxxxxxxx> wrote:
> is the consensus that RAC is too immature for primetime use? What about
> other non-linux unix platforms - aix, solaris maybe?
I don't think that the consensus is that RAC is not ready for
primetime use, that isn't my understanding of the various discussions
that I have seen around the subject. My understanding is that it adds
an extra layer of complexity and takes some extra care and feeding as
compared to a single instance database. (That's why the consultancies
will gain - at least for the next year or two - when RAC gets
installed. ) The benefits you get are the scalability, and to a
certain extent availability that comes with it - together with - if
you are moving from proprietory Unix hardware to el-cheapo intel/linux
servers reduced hardware costs.
Given your explanation of the criticality of your apps, it may well be
that RAC does give you added availability - fail-over to an additional
node in the event of instance/node failure will happen with less
downtime and for fewer clients than failover to standby. I'd expect
that it won't happen with *no* downtime for affected clients unless
your apps are a) suitable for and b) written for TAF, so I'd be
looking to test what happens to a client which is connected to a node
that fails and what impact does that have on patient care.
I'd also still think about the single database as a point of failure.
Carel-Jan's points about speccing out what downtime is acceptable and
what it will cost to assure it are well made.
--
Niall Litchfield
Oracle DBA
http://www.niall.litchfield.dial.pipex.com
--
//www.freelists.org/webpage/oracle-l
Other related posts:
