[oracle-l] Re: is this a failover - or ....
- From: Carel-Jan Engel <cjpengel.dbalert@xxxxxxxxx>
- To: oracle-l@xxxxxxxxxxxxx
- Date: Tue, 27 Jan 2004 19:42:07 +0200
Markus,
You described an almost perfect failover, and far cheaper than RAC.
Actually, we set this up years ath the datacenter of a local government.
They had many applications, and appr. 18 oracle instances running.
Two HP servers, one storage. Half of the databases mounted form one server,
the other half from the other. In case of a failure, all non-critical apps
were stopped, and the critical ones were started from the remaining server.
I know of a dutch bank, that is setting up their HA systems in the way you
describe, thus saving from the horrendous costs of RAC. When you don't need
the scalability, and can afford a few minutes downtime, your're safe. They
call it PM/RAC, Poor Man's RAC. As a bank, they're always short of money,
that's why. :-)
There is at least one Single Point of Failure remaining: your storage
cabinet. No problem, but take care of proper procedures what to do if the
storage fails. You can mirror that to another storage cabinet (or SAN, NAS,
or whatever non-BAARF-banned configuration you can think of (see
www.baarf.com)) or consider using Data Guard, one of my favourites ;-).
Regards, Carel-Jan Engel
===
If you think education is expensive, try ignorance. (Derek Bok)
===
At 06:45 PM 1/27/2004, you wrote:
>we have - amongst others - 2 sun fire v240, each has fibre channel =
>adapters and is connected to a storage system. a storage system is a =
>system, that makes the connected machine(s) - if correctly configured - =
>believe, that there is one ( or more ) additional scsi drive(s), that can =
>be treated like physically built in device(s). in this document i'll call =
>it vdisk (like HP does).
><gggg>
>(in case you all know it: forgiveness for beeing smart allecky)=20
><gggg>
>
>tricky: both of the 2 sun-v-240 can mount both devices, so - in a =
>completely stupid and unusable case - both machines have both devices =
>mounted in rw. never considered this as a seriously working configuration. =
>*BUT* if one sun-v-240 is down, the other one can mount it in rw and use =
>it. this is a working - and not stupid - situation. (admitted: the =
>sun-v-240 is not likely to be down, but just for the sake of this now let =
>us assume) =20
>
>consider the scenario described the paragraph before. and now consider =
>there is an (identical) oracle installation on both machines, correctly =
>designed (pathes, sysmlinks...), so that both machines can - mutually =
>exclusively - mount the device and start the oracle machine (instance) and =
>mount and open ... the database(s). it works. listeners can be configured. =
>works too.=20
>
>can this be considered a failover scenario???=20
>did I possibly overlook some nasty detail???
>
>any input appreciated. (... even rebukes)
>
>assumed: the storage area network (2TB) never really fails - raid 5, .... =
>and so forth.
>
>kr MR
>
>-------------------------------------------------------------
>Please see the official ORACLE-L FAQ: http://www.orafaq.com
>-------------------------------------------------------------
>To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
>put 'unsubscribe' in the subject line.
>-------------------------------------------------------------
-------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
-------------------------------------------------------------
Other related posts:
