Re: Can O_SYNC be implemented by using fsync?

From: Jeff V. Merkey (jmerkey@timpanogas.com)
Date: Wed May 17 2000 - 09:10:48 EST


I guess this is OK if Oracle and other DB's are always restricted to
local file system usage (since using direct block interfaces). Having
all the FS's able to support Oracle data sets might be a desireable
feature in the future ...

Jeff

Andreas Dilger wrote:
>
> Jeff V. Merkey writes:
> > Also, Van mentioned the issue with double-caching. Since Oracle uses
> > it's own cache, the current buffer cache implementation in Linux for
> > O_SYNC obsolutley sucks for Oracle, since we will always be double
> > buffering data. I mentioned to Alan having a directFS layer in Linux,
> > where Oracle could talk to files in a way that would completely by-pass
> > the Linux buffer cache. Given the current state of Linux and the file
> > systems, this may be very difficult to support.
>
> What about the raw-io support? If you are using a filesystem, you have
> to live with how a filesystem works. If you really need direct device
> access, you can use raw device access, raw-io (to avoid buffering) and
> LVM (to avoid the partition nightmare).
>
> Dan Kegel writes:
> > Say, I thought Oracle could use raw disk access instead
> > of going through the filesystem. Is O_SYNC used only
> > when in a 'convenience development mode' where you are
> > using the filesystem, before you go to the trouble of
> > giving Oracle a whole raw partition?
>
> Some people run with databases in files, but that is for people who don't
> care about performance, IMHO. Double cache is one performance problem,
> but there is also overhead from the filesystem itself (e.g. indirect block
> lookups, atimes, no guarantee of linear file allocation in filesystem, etc).
> Raw partitions are the only way to go, but really need LVM and raw-io in
> order to make this useful.
>
> Cheers, Andreas
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Tue May 23 2000 - 21:00:13 EST