RE: intel clusters in a box

From: Marquez, Chris <CMarquez_at_aarp.org>
Date: Wed, 26 Jan 2005 17:04:23 -0500
Message-ID: <7E412C164E6ECB468834A39F31E6E0D406689015@mbs06dc.na.aarp.int>

>> RAC is one database with multiple nodes
>> if you have two disk arrays,
>> how can it be one database?=20

It can't...I mean I guess HP would argue that this is the same "logical database",=20
but I'm sure that there are *two* "/../system.dbf" datafiles at "two" different data centers!?
and that is not one "physical database"...

Chris Marquez
Oracle DBA
HEYMONitor(tm) - heymonitor.com
"Oracle Monitoring & Alerting Solution"

-----Original Message-----
From: Paul Baumgartel [mailto:paul.baumgartel_at_gmail.com]=20 Sent: Wednesday, January 26, 2005 4:24 PM To: Marquez, Chris
Cc: oracle-l_at_freelists.org
Subject: Re: intel clusters in a box

Chris,

Yes, I am also skeptical of Extended Cluster for RAC, and the thing that trips me up is the software replication. If it's true RAC, it's one database with multiple nodes; if you have two disk arrays, with replication (mirroring) between two arrays involved, how can it be one database? I just don't see how that works.

PB

On Wed, 26 Jan 2005 13:11:31 -0500, Marquez, Chris <CMarquez_at_aarp.org> wrote:

>=20

> Paul,

>=20

> Sorry for the slow reply.

>=20


> >> http://h71028.www7.hp.com/enterprise/cache/10084-0-0-225-121.aspx

> >> The Extended Cluster for RAC is...
> >> a virtualized application environment
> Not sure what this means.

>=20

> >> a replicated and synchronized database.
> "replicated"...This simply scares me... :O|

>=20

> >>Extended Cluster for RAC works in an active-active mode
> Would have to see it action before buying in.

>=20

> >>across two datacenters, up to 100 kms apart
> 100 kms =3D 62 miles...nice but as I said before this (SAN syncing) =
has=20
> geographic limitations. My clients have data centers in different US=20
> states if the have one at all).

>=20

> >>HP worked closely with Oracle, AT&T and Nortel Networks
> Yeah...got to have a great network to before even talking.

>=20

> >> http://h71028.www7.hp.com/enterprise/downloads/SgeRAC_wp_030504.pdf
> There is a nice graphic on page 4 of the .pdf

>=20

> Also on page 5;
> >> "Database replication is achieved via host--based software=20
> >> mirroring, with RAC synchronizing the database caches via Cache=20
> >> Fusion."

>=20

> BUT what happens when these to "physical differ database (being
> replicated) become out of sync!?
> Sorry but I'm no fan of "software based replication"...this is counter

> the concept of OPS/RAC...which his based on hardware=20
> replication/redundancy/sharing!

>=20

> If you have sat and watched your (v$...) sessions waiting for "global=20
> cache..." with two RAC nodes side by side, I can only imagine the fun=20
> it would be with two RAC nodes in different data centers.

>=20

> This all sounds good, but simply, "why"?
> Nice to see, but honestly I would never want to mange this...I'll take

> DataGuard for now.

>=20

> I don't want to sound cynical...this is very cool and good info...I=20
> appreciate it. Oracle OPS has come a long way.

>=20

> Chris Marquez
> Oracle DBA
> HEYMONitor(tm) - heymonitor.com
> "Oracle Monitoring & Alerting Solution"

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

Received on Wed Jan 26 2005 - 17:07:56 CST