Re: async i/o via nfs?

"Sybrand Bakker" <gooiditweg_at_sybrandb.demon.nl> schrieb im Newsbeitrag news:053viv4ijhiguqjaf7kke0io8339hmliem_at_4ax.com...
> On Tue, 5 Aug 2003 11:34:10 +0200, "Volker Hetzer"
> <volker.hetzer_at_ieee.org> wrote:
>
> >The problem is
> >that one of my controlfile copies and one of my archivelog copies
> >sits on another machine that is mounted via nfs.
>
> Shudder. How on earth did you decide to locate one of your controlfile
> copies on a nfs-mounted drive? Why not all of them?
I've got three copies, one near the db files, one on another volume and one on another machine. So far it has worked perfectly. What would you suggest for having a copy on another machine?

> nfs is a can of worms, it is slow, it is undependable, and Oracle
> advises expressly against using it. I would try to use async i/o if I
> was looking for a different job in the near future, but not in other
> circumstances.

I'm trying it on a test system anyway, so don't worry. :-)

> Try it, and follow the results by means of truss
Where does oracle advises this?
I've just had a look and what I got was "RAC on Linux Best Practices" and it says Asynchronous I/O can be used on whatever and nfs too.

> My bet is your async i/o will fail and Oracle will resort to normal
> file i/o anyway. Though not once, but for every individual request.
Is there any way to check this?
Or, rather is there any way to check reliably whether async i/o is used by the db?

> Also, if Linux uses ufs (unix file system) async i/o won't work
> anyway.

Well, I use the redhat es because of kernel async i/o. What is ufs async i/o?

Lots of Greetings and thanks!
Volker Received on Tue Aug 05 2003 - 06:38:36 CDT