Re: ASM of any significant value when switching to Direct NFS / NetApp / non-RAC?
Date: Fri, 10 Aug 2012 08:40:22 -0600
Message-ID: <50251D56.7020505_at_gmail.com>
On 10/08/2012 4:24 AM, Dana Nibby wrote:
> _at_Hans, interesting about the issue of HW mirroring versus SW multiplexing and how the two differ. Is Alex's demo available online? If not, can you recommend any good sources discussing the differences / perils in more detail? While I'm no longer "pushing" for ASM (I'd call it due dilligence / inquiry / guarding against the potential Loss of a Good Thing without sufficient evidence) I'm still interested in knowing the subtleties around these matters.
HW mirror is an exact duplicate, down to the removal using an accidental (or malicious) "rm *". Easy enough to demonstrate - create DB with one control file (because it's h/w mirrored, no need for 2, right?); blow it away; restart your DB. Do same thing, but create 2 control files *in separate locations and on separate disks*. Compare recovery time.
A general, dated, article about disk corruption and how often it
happens:
http://www.zdnet.com/blog/storage/data-corruption-is-worse-than-you-know/191
In general, data corruption on disk happens often enough that it's
something to think about. And - more importantly - it's often enough
that it could impact the data set (tablespace) several times while
waiting for SAN to replace/rebuild a specific protected disk.
My concern is simple - most SAN administrators think in terms of many small files. Corrupting one file, and 'oh well'. Database has few, very large, files. Corrupt one file, and 'oh oh' unless there are suitable protections.
You'll need to talk to Alex to see whether he's posted the slides and demo somewhere. I know that he'd planned to post it with the Collaborate 2012 materials, but I haven't checked. He works for Pythian - easy enough to look up. Or come to Oracle Open World and corral him there.
/Hans
-- http://www.freelists.org/webpage/oracle-lReceived on Fri Aug 10 2012 - 09:40:22 CDT