Re: Is Raid 5 really that bad for Oracle?

Daniel Morgan <damorgan_at_x.washington.edu> wrote in message news:<1091677845.282907_at_yasure>...
> Noons wrote:
>
> > Daniel Morgan apparently said,on my timestamp of 4/08/2004 12:22 AM:
> >
> >>>
> >>> Wait until the IT damager decides to go to a NAS
> >>> because it's "better value"...
> >>
> >>
> >>
> >> The performance difference between SAN and NAS used to be of critical
> >> importance. I am not seeing enough difference these days to justify
> >> the huge difference in price. Is anyone having a different experience?
> >>
> >
> > Of course there are exceptions. But IME, I've never seen a
> > network-based I/O system that can provide the performance
> > scalability of a channel I/O system. Don't forget that
> > a NAS uses one or more network connections presumably in a
> > single card.
> > Now I don't care how fast your Ethernet may be, it just by design
> > cannot cope with a busy I/O demand. Single-task, single-user
> > (the way "scalability is "measured" nowadays...) it may
> > show good results. Wait for the first collision...
>
> Actually we use ethernet binding ... so we pretty easily construct a
> far bigger pipe.
>
> Not fiber I'll grant you but still ... the difference doesn't seem to
> be large enough to justify the cost.

Daniel,

"Ethernet binding"?
Sounds like your networking config is not getting enough fibre. ;) I prefer trunking myself, but then again, its vendorspeak.

"aggregation" might be vendor agnostic.

ok. this was a troll in that it provided no real value other than an attempt at a cheap laugh in that Storage mounted over Ethernet mind have digestive track problems.

back to the front ...

-bdbafh Received on Thu Aug 05 2004 - 09:59:29 CDT