RE: RAC on extended distance clusteres

$$$$$$ - the *dark fibre* comes to mind here in the states to do something like this ...

Greg

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Bryan Thomas Sent: Tuesday, December 13, 2005 1:43 PM To: peter.sharman_at_oracle.com
Cc: stellr_at_cns.vt.edu; oracle-l
Subject: RE: RAC on extended distance clusteres

Here is a quote from an internal Oracle Metalink article:

Real Applications Clusters (RAC) on Extended Distance Clusters is an architecture that provides fast sub-minute recovery from a site failure and
allows for all nodes, at all sites, to be part of single database cluster.
These advantages have raised interest in the architecture, have led some customers to implement, but it is critical to understand where this architecture fits best especially in regards to distance/latency and degree of
protection, as well as some of the additional implementation complexity.

The high impact of latency, and thus distance, do create some practical limitations as to where this architecture can be deployed. This architecture
fits best where the 2 datacenters are relatively close (<~25km) and where the
extremely expensive costs of setting up direct cables with dedicated channels
between the sites has already been taken. All of the known customers implemented with distances less than 25km, and because of the performance
impact as the distance is increased it is unlikely they will be implementations
at distances greater than 25km.

RAC on Extended Distance Clusters does provide a greater HA than local RAC but
may not fit the full Disaster Recovery requirements desired. 25km is great
protection for some disasters (local power outage, airplane crash, server room
flooding) but not all. Disasters such as earthquakes, hurricanes, and regional
floods will affect a greater area. Customers should do an appropriate analysis
to determine if both sites are likely to be affected by the same disaster.

RAC and Data Guard combined does provide more comprehensive protections not
covered by RAC on Extended Distance Clusters alone. This includes protection
against corruptions, users errors (in 9i), protection against regional disasters, and support for full rolling upgrades across Oracle versions.

Configuring an extended distance cluster is more complex than a local cluster.
Specific focus needs to go into node layout, quorum disks and data disk placement.

Bryan Thomas
Senior Performance Consultant
Performance Tuning Corporation
www.perftuning.com
(512)751-5516
bthomas_at_perftuning.com
Quoting Pete Sharman <peter.sharman_at_oracle.com>:

> Well, you got me there, it should have been "fell" and I'd never heard
of
> Paddy Martin, so that makes us even! :)
>
> I'm not saying it won't work, it will and we have customers doing it.
But to
> me it's trying to fit a square peg into a round hole. Use a big
enough
> sledgehammer and you can do anything I guess. However, if your data
is that
> important to you that you want to make it highly available, then use
the
> right technology for the right problem. HA is hard enough as it is
without
> doing weird and wonderful things to achieve it. Just using RAC with
extended
> clustering technology does not give you the full gamut of HA that you
can get
> using RAC for machine failover capability and DataGuard for site
failover
> capability.
>
>
> Pete
>
> "Controlling developers is like herding cats."
> Kevin Loney, Oracle DBA Handbook
>
> "Oh no, it's not. It's much harder than that!"
> Bruce Pihlamae, long-term Oracle DBA
>
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org

[mailto:oracle-l-bounce_at_freelists.org] On
> Behalf Of Ray Stell
> Sent: Wednesday, 14 December 2005 4:56 AM
> To: Pete Sharman
> Cc: oracle-l
> Subject: Re: RAC on extended distance clusteres
>
> On Wed, Dec 14, 2005 at 04:09:07AM +1100, Pete Sharman wrote:
> > It seems people try to use this as a cheap way to provide DR and HA
in one
> foul swoop.
>
> Forgive the non-techie aside, I had never heard the foul ref before,
> though it is reported to be in common usage. So, what you're saying
is
> that RAC for DR purposes is a "Paddy Martin?"
>
> http://www.worldwidewords.org/qa/qa-fel1.htm
>
> "All my pretty ones?
> Did you say all? O hell-kite! All?
> What, all my pretty chickens and their dam
> At one fell swoop?"
>

--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l