> Pavel,
>
> The architecture of the RAID drivers is extremely primitive, and
> limiting, and does not handle hotixing or distributed cases. There are
> dependencies on NWFS for controlling th mirroring. At present, the
> buffer cache does not support a logical semantic. It's true we could
> "stand on our heads" and map something very unnatural to what's there,
> but there are still issues and problems with the buffer cache and how it
> acts. There are also config issues. NWFS style allows end users to
> configure and use mirroring on Linux without needing a degree in
> astro-physics to understand how to set it up.
>
> We basically need a smarter cache than what's there. Eventually, Linux
> will evolve there (Since the page cache already resembles what's in NT).
Your statements look a bit unclear to me.
1) You seem to state that NWFS is controlling mirroring from the FS. If
true, this is different from the way Linux handles RAID.
2) On the other hand you claimed that handling some list of physical
devices to mirror a logical device at buffer cache level (NT) is far
superior than the way Linux handles RAID.
I can understand that (1) perhaps allows to implement some features not
possible with the layering used in Linux, at the price of increasing FS
complexity probably.
About point (2) I donnot see differences in essence, given your
description of the way NT handle caching.
I am always amazed by "logical .vs. physical" considerations that
generally just mean that some layer or abstraction has been added
somewhere, and nothing more.
A single invention of the wheel has been enough for the real physical
level, but this process seems to be recursive in information technology.
;-)
Gérard.
> Jeff
>
>
> Pavel Machek wrote:
> >
> > Hi!
> >
> > > Which is exactly what we have concluded. We probably should provide
> > > some input into 2.5 for our needs. We would be happy to put in the
> > > subsystems we require. The benefit to Linux would be a built in fault
> > > tolerant redirector within the cache (which is useful for this and
> > > distributed support -- time to start thinking about this). This would
> > > allow all of the Linux file systems to support multi-segmented
> > > mirroring and fault tolerant failover without the RAID drivers.
> >
> > What is problem with the RAID drivers?
> > Pavel
> > --
> > I'm really pavel@ucw.cz. Look at http://195.113.31.123/~pavel. Pavel
> > Hi! I'm a .signature virus! Copy me into your ~/.signature, please!
> >
> > -
> > 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/
>
>
-
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/