Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Usenet -> c.d.o.server -> Re: Why?
"Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:<40e00319$0$18666$afc38c87_at_news.optusnet.com.au>...
> "Alexander Skwar" <from_at_alexander.skwar.name> wrote in message
> news:514cqrts7o0q.dlg_at_a.skwar.digitalprojects.com...
> > Am Mon, 28 Jun 2004 19:48:03 +1000 schrieb Howard J. Rogers:
> >
> > > "Alexander Skwar" <from_at_alexander.skwar.name> wrote in message
> > > news:8iowpa5bgf2f.dlg_at_a.skwar.digitalprojects.com...
> > >> Hi!
> > >>
> > >> In the Oracle documentation, it's often suggested to use OFA to design
> > >> where files are located. This suggests to create filesystems /u00,
> /u01,
> > >> /u02 ...
> > >>
> > >> What I don't get - suppose that my server supports RAID 5.
> > >
> > > No database should run RAID5.
> >
> > Why?
>
> Because RAID5 has a write penalty (so, Ok, a read-only database is probably
> OK on RAID5).
>
> >And why are you even suggesting RAID0? While RAID0 surely improves
> > the speed, it also increases the risk of losing everything when just one
> > disk goes faulty.
>
> What are archivelogs there for?
>
> My point was simply that RAID0 stripes, but without the write penalty.
> Practically, you would really want RAID0+1, of course.
>
> But that's expensive, and one can forego the +1 component with a clear
> conscience, provided your backup and archiving strategies are well thought
> out.
>
> Regards
> HJR
HJR, you are really dumb!
Oracle data scenters use Network Appliance filers that are basically built with RAID4 technology - similar to RAID5.
Please do not generalize! Most RAID boxes use large caches and do the writing using xx-stage algorithms your alikes are not fit to understand.
to mirror; to stripe; to stripe mirrors; to mirror stripes; I bet your expertise with these are as shallow as your knowledge of access methods, optimizers and/or plans.
Apardon my response DOC! But please realize your reply is full of .... Received on Mon Jun 28 2004 - 14:40:16 CDT
![]() |
![]() |