[Info-vax] RAID vs. MOUNT/BIND

[Ken Fairfield]

On Nov 23, 11:25 am, Rob Brown <mylastn... at gmcl.com> wrote:
> On Tue, 23 Nov 2010 at 09:15 -0800, Keith Lewis wrote:
> > ... RAID 5 should theoretically give you both performance and
> > redundancy.
>
> RAID 5 gives redundancy, no argument there.
>
> RAID 5 gives good read performance, but its write performance is not
> as good.  Each logical write becomes 4 I/O: 2 reads and 2 writes.
>
> If all of the extra I/O is hidden in hardware with deep caches, then
> it may not be noticeable.  Or if you don't do so many writes, you may
> not be bothered by poor write performance.
>
> Or at least, that is the way it seems to me.
>
> PS:  *Is* there a host-based RAID (other than 1 and 0) for VMS?

Yes, there *is* host-based raid for VMS.  In fact, I think
that's it's product name. :-)

We used it when I was at my former employer.  It's
pretty tightly integrated, and depends upon, host
based volume shadowing.  My recollection is that
you have only the single choice of RAID 0+1, i.e.,
striped shadow sets.

It worked well, but could be tedious to set up
initially.   A failed shadow set member was
handled pretty much in the usual way, but ISTR
the product be "happiest" if you used the RAID
commands  to add a 3rd member to copy and
to subsequently drop a failing member, etc.

I don't recall having any failures with the product...

   -Ken