Re: NT striping dangerous with Oracle 7 or 8?

What about the fiability of a commit statement ? IMO, the commit statement should take care of physical write on disk. What you said
sounds like "striping driver says to process that data is wrote, but it is not really wrote".
When writing on disk, it is not a matter of commit response quickness, but sure the
trustness of a driver's response.
I think that you talk about write caching, rather than disk striping. And this is the same on any OS, isn't it ? Couldn't the write caching of the striping driver be turned off ?

MK :-)

Uday Bikkasani wrote:

> NT striping is always dangerous. Try the Raid striping at the hardware
> level. Also make sure that you have battery backups on the raid controllers.
>
> If power is lost there are many points of failure even if make sure that you
> have battery backups. Make sure you have a good backup strategy implemented.
>
> uday
>
> bdurrett_at_ccci.org wrote in message <6d6l3i$ema$1_at_nnrp1.dejanews.com>...
> >I have been told that it is dangerous to use NT striping with Oracle on NT
> >4.0. This person told us that NT buffers some Oracle writes in memory and
 if
> >the server is turned off at the wrong time the data could be corrupted. I
> >find this hard to believe, but I wanted to post this question just to be
 sure.
> >Is there a problem using NT striping and oracle 7 or 8 server?
> >
> >Thanks,
> >Bobby Durrett
> >
> >-----== Posted via Deja News, The Leader in Internet Discussion ==-----
> >http://www.dejanews.com/ Now offering spam-free web-based newsreading
Received on Mon Mar 02 1998 - 00:00:00 CST