Re: ASM of any significant value when switching to Direct NFS / NetApp / non-RAC?
- From: Hans Forbrich <fuzzy.graybeard@xxxxxxxxx>
- To: oracle-l@xxxxxxxxxxxxx
- Date: Thu, 09 Aug 2012 09:11:24 -0600
Like any other technology, ASM is a tool.
Last time I watched a *professional* carpenter, he had several hammers
in his tool belt. While there were overlaps in what could be done,
each hammer also had a special purpose, and when used by a professional
for that special purpose, the job *could* get done faster/cheaper/better
than by using a general purpose tool.
But specialty tools, and specialty technologies, can also become the
basis for technology religions, and religious wars.
When compared to raw device management, I certainly prefer ASM. If a
shop insists on using SANs, and if the shop does not have experience in
cluster file systems, it has significant advantages.
And the idea of leaving RAID 5 behind appeals to me - especially with
the silent corruption problem experienced on some modern SANs. So if my
storage admin team tells me that I am stuck with RAID 5 disks,
configured and allocated based on experience with MS operating systems;
and then my *nix adminbistrator tells me he wants to add a layer of PVM
and LVM management, configured and allocated based on ext3 experience
(lots of small files, rather than few large files) - I'd want to be
thinking of using a different tool.
But if my storage, network and OS support guys - who will point the
finger at the DBA if anything goes wrong, until the DBA proves that it
is not a DB issue - are capable of handling a decent large scale,
database oriented, network file system from NetApp or ZFS Appliance,
then I'm pretty comfortable with that direction.
Just don't tell me that hardware mirroring is exactly the same as
software multiplexing. At that point, cred is shot - "rm *.ctl" for
example.
I like Alex Gorbachev's demo at Collab 2012, where he proved (to my
satisfaction) that loss of several disks in a mirrored situation can
lose/corrupt data, whereas Normal and External redundancy ASM will stop
the database before data loss/corruption is possible.
/Hans
On 09/08/2012 7:07 AM, Dana Nibby wrote:
> Anything that reduces politics typically increases efficiency and
> effectiveness. Politics, while I suppose they are necessary--it's the way
> some people Get Things Done (and so can't be completely avoided)--seem to me
> largely to be a drain on brainpower and lead to inefficient deployment of
> staff resources. The goal should always be to increase efficiency and
> customer satisfaction. And not to Build Personal "Empires" and engage in
> other hoarding/face-saving activities. I don't believe IT is a zero sum game.
> It is, or ought to be, a Team Sport (and I am no fan of sports; but it's a
> helpful analogy).
>
> If "no ASM" is the right way to go for the best interests of the customer,
> I'm 100% on-board withat that. I'm of the perhaps naive opinion that, in
> environments where there are many levels of separation of duty, that we all
> work on the same team. There is, or should not be, an "Us versus Them"
> mentality. The correct question to ask is nearly always this: what
> constellation of technologies is best for downstream customer productivity.
> And "if that's wrong, I don't wanna be right." :-)
>
> Best,
>
> Dana
--
//www.freelists.org/webpage/oracle-l
Other related posts:
