Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Direct IO

Re: Direct IO

From: Dave Haas <davidh_at_no.spam.hotmail.com>
Date: Sat, 23 Jun 2001 23:49:27 GMT
Message-ID: <bW9Z6.2927$84.465091@news0.telusplanet.net>

Hi Dave.

Is it possible to send IO requests through the Unix file system layer and ask it not to buffer? My understanding was the two methods are:

  1. File system IO (which means the IO request is sent to the device through a standard Unix IO system call, in which case the file system in use gets to decide whether to cache and buffer IO requests) and
  2. Raw IO (which means that Oracle essentially gets to go around the file system layer and talk to the device driver directly, i.e. opening the /dev/hdaXX block device directly and issuing your own read_x_bytes_starting_here() and write_x_bytes_starting_there() calls).

Is there something i'm missing here? How does one issue a IO request and not have the file system decide how to do the buffering (because I imagine each file system does buffering differently ...)

Cheers,

Dave

"Dave Grantier" <dave_at_foobar.com> wrote in message news:3b33f176.1305136_at_news.mindspring.com...
> David,
> Here's my two cents.
> I understand direct IO to mean something to the effect of bypassing
> the Unix Buffer cache. Some Unix filesystems allow you to disable
> caching. I'm thinking of HP's JFS (by Veritas), and, of course,
> Veritas.
> Naturally, Raw IO bypasses the whole schmeer.
> Cheers,
>
> On Sat, 23 Jun 2001 00:45:38 GMT, "Dave Haas"
> <davidh_at_no.spam.hotmail.com> wrote:
>
> > Hi.
> >
> > I have a question with regard to terminology. In several posts (and an
> > argument I had with Howard a while ago :) people have used the term
 'Direct
> > IO'. To be perfectly honest I'm not exactly sure what that means.
 AFAIK
> > the IO options are 1) file-system buffered and 2) Raw IO. I heard (or
 more
> > to the point, read) a post that said something to the effect of '...
 direct
> > I/O means that the buffer cache is not involved in the operation ...'.
 Does
> > that have something to do with the sort-direct-write operation and the
 old
> > SORT_WRITE_BUFFERS and SORT_WRITE_BUFFER_SIZE? I'm a little confused
 (as
> > usual) ...
> >
> > Dave
> >
> >
>
> Dave
>
> +---------------------+---------------+
> |nojunkmail_at_nospam.com| bellsouth.net |
> |nospam_at_nojunkmail.com| davegrantier@ |
> +---------------------+---------------+
Received on Sat Jun 23 2001 - 18:49:27 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US