Re: Hardware RAID vs. Software RAID

"Chuck Lucas" <chuck.lucas_at_mspb.gov> wrote in message news:bn15d0$5t7$1_at_ngspool-d02.news.aol.com...
> "Daniel Morgan" <damorgan_at_x.washington.edu> wrote in message
> news:1066587431.204277_at_yasure...
> > You are correct on all points.
> >
> > But what version you implement can make a big difference. I'd stay away
> > from RAID 5 and vendor RAID
> > implementations with numbers like 2 and 4.
>
> What's wrong with RAID 5?

Well here is where I disagree with many folk - Daniel included and also pertinently myself from a couple of years ago. The fact used when discussing RAID5 for database work is its absolutely terrible performance when writing (actually and some reads as well). This is all true and simple physics. The question for me is - is this penalty a problem? For me the answer would be yes *every single time your system is or might become* IO bound. In many cases however there are other limitations (CPU being the prime example) and the physical IO can easily be handled by a reasonably large Raid5 array. Moreover even when the system is doing too much physical IO for the disks/controllers its a dime a dozen that its actually just doing too much IO (physical and logical) - that will hurt you everywhere.

Now obviously if you have a system where it really is the speed of the disks or the MHz rating of the CPU that is holding you back then fine go RAID10 you know it makes sense (and is undoubtedly *the Right Thing(tm)* ) otherwise address your actual problems first of all.

For the technical discussions about RAID5 and why it is poor compared to other stuff go to www.baarf.com

--
Niall Litchfield
Oracle DBA
Audit Commission UK
*****************************************
Please include version and platform
and SQL where applicable
It makes life easier and increases the
likelihood of a good answer
******************************************