Java Mailing List Archive

http://www.dba.5341.com/

Home » Home (12/2007) » suse oracle »

Re: [suse-oracle] Server and Service Level Failover

Rami

2006-04-11

Replies:

C'est Pierre wrote:

> I am currently entitled to evaluate which solutions can support an
> active/passive ("Cold Failover") system based on SuSE Enterprise Linux
> 9 running Oracle 10G.
> On another SuSE maillinglist, I was told about Oracle RAC and I talked
> to our local Oracle DBA. He told me that RAC serves the purpose when
> there's 2 oracle instances, each on a different server, both sharing
> the same storage. That's not our purpose. We want that if the active
> node (server/instance) dies for some reason (network outage, hardware
> failure, etc) the other server takes the lead, launches an oracle
> instance and mounts the storage.
>
> We are going to use the hardware on the following list to accomplish this:
>
> 2x HP BL25p, with 4xOpteron 200 series and 8gb ram. Fibre channel adapter
> 1x HP StorageWorks EVA 3000 for shared storage.

Hi,

Well, Polyserve does that with NAS. But you should have two NAS'es. The
material you described is perfect for two-node RAC with shared storage.
On the other hand, if you've done some mountaneering or caving, you
should ask yourself: "What happens if the shared storage goes down? Do I
fall and die?"

The cheapest and least fault-prone solution is DataGuard: two servers on
your LAN in MAXIMUM PROTECTION or MAXIMUM AVAILABILITY mode, eventually
a third remote one if you've got a 10 Meg WAN connection in MAXIMUM
PERFORMANCE mode to ensure backup in case of nuclear war.
With DataGuard, you will need only one Oracle licence, unless you use
the standby server for reporting purposes over a few days a year.

The downside of DataGuard is that you will need to issue three or four
commands if your main server goes down. No automatic startup.

You might also have a two-node RAC configuration with shared storage as
you described, plus one DataGuard standby server to start when both of
your RAC servers panic simultaneously because a mouse was running on the
server room's floor.

Basically, your purpose is to ensure maximum data protection and minimum
downtime. Not increasing downtime because of a technology you don't
master, or loosing your data because you put all of your data in one place.

Rami, who's seen seven disks all hosed up in a shared storage bay.


--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting

©2008 dba.5341.com - Jax Systems, LLC, U.S.A.