RE: Cluster file systems versus raw devices in Oracle RAC
- From: "Kevin Closson" <kevinc@xxxxxxxxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 30 Dec 2005 10:32:03 -0800
Have you seen/implemented EFSCG for Oracle databases ?
Ahhhh...but of course. Me testing with Oracle? :)
Oracle is examining our OSCP testing results now. Things look
good. I'll be presenting a Proof of Concept paper at RMOUG
Training days in Feb.
Let's just say that a really busy table (or write intesive anything) in
any other OSCP NAS out there can easily become the bottleneck.
Instead of getting another filer, pulling some of the hot data out and
plunking it into the new filer, you just add a NAS head in this
architecture and serve up the same files via that new head along with
whatever
heads were previously serving up those files (scalability without
replication)
...data stays in place. Oh, also, I can power off an active NAS head
right
our from underneath an PQO index creation and nothing skips a beat...now
that is real
HiAv storage for RAC. I can power off the NAS head that is serving up
the
CRS files too without any impact to Oracle.
I would go into the particulars of why EFS.CG is good for
Oracle deployments, but there is a dangerously high percentage
of list participants here that view anything other than "the way we
have always done it" as some sort of personal insult.
<http://web.singnet.com.sg/~hkchital>
Other related posts:
