The raw device "restriction" is an OS thing - Unix can't share files without
it, until you get to some of the new clustered file systems (available in
Sun Cluster 3.0 and some others as well). If it was on good ol' VMS, you
wouldn't need raw devices, but then again VMS is a real operating system.
;)
I shall sit back and wait for the flames!
--
HTH. Additions and corrections welcome.
Pete
Author of "Oracle8i: Architecture and Administration Exam Cram"
Now got a life back again that the book is released!
"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
"Howard J. Rogers" <howardjr_at_www.com> wrote in message
news:3b953977_at_news.iprimus.com.au...
> It's actually called Real Application Clusters (RAC for short).
>
> You have multiple Instances running on separate, but interconnected, nodes
> (machines). They each access the one set of shared disks (which from
memory
> have to be raw devices). Each node can also have its own set of private
> disks, for storing local copies of the Oracle executables, and
(preferably)
> the locally-produced archive logs.
>
> Availability is sky-high, because any number (except all of them!) of
nodes
> can simply fail, and yet your Users can quickly re-connect to the
surviving
> Instance(s). It's even better than that, though, since there is also Real
> Application Clusters Guard built in (an included extra that resembles the
> old Parallel Fail Safe) to perform automatic failover in the event of an
> Instance failing.
>
> Regards
> HJR
>
>
>
>
> <f_dubru_at_yahoo.com> wrote in message
> news:20010904.12115152_at_mis.configured.host...
> Hello,
>
> I am having some information about Oracle 9i before planning a possible
> move to this DB. I am particularly interested in the Real Application
> Server but still have a couple of questions about availability.
>
> After what I read, I understand that each node in a cluster has full data
> access making the system quite safe as long as at least one node is up
> and running. I guess by node, I must understand a processing node and
> that I still need a replicating data system (such as RAID5 or Legato) for
> the database files for full fault resilience. Is that correct?
>
> Thanks a lot for your time.
>
> Fred.
>
>
Received on Tue Sep 04 2001 - 18:42:50 CDT
