There are additional considerations with using OPS the biggest of which
being the application tenability to OPS. If the application is not suited to
OPS, the instances will throttle each other with inter-instance locking
requests and 'false pinging' of data blocks. You can read more about it in
the Parallel Server admin guide.
OPS option needs to be purchaed separately ( per user licensing fee ),
you'll also need to purchase DLM/cluster mgmt software.
Jay Katti <mkatti_at_hns.com> wrote in message
news:7o6uc5$j6c$1_at_hnssysb.md.hns.com...
> Greetings to all.
>
> While OPS provides fault-tolerance and availability with multiple
instances
> there is still just a single point of failure (with the RAID disk array)
> when used
> with a cluster. So I plan to mix OPS with a standby database.
> I prefer OPS over replication since my sites are not too far
> apart and the failover is transparent to my applications. (Whereas in
> replication,
> I may need to deal with deferred transactions, replication groups and
more.)
> Any gotchas and hurdles with using OPS ? How much does a cluster
> with two nodes on a Solaris box cost for use with OPS ?
>
> Do you have any suggestions on mixing OPS with an additional standby
> database, so that in the worst case that the RAID fails, we can have bring
> up
> the standby database ?
>
> I am (still) using Oracle 7.3.4, so I would not be able to use the Oracle
8i
> features such as 'automated standby'. But I can still have the
> 'redo log' files transferred often to the fail-over database in 7.3.4.
>
> Any suggestions, comments or advice would be appreciated.
>
> Thank you.
>
>
> Jay ...
>
>
Received on Sat Aug 07 1999 - 14:51:09 CDT
