Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Usenet -> c.d.o.server -> Re: raid level
On 15 Dec 2005 22:36:16 -0800, "ORA600" <panandrao_at_gmail.com> wrote:
>Hi,
>
>As much as i would like to agree with all on this post about RAID 0+1
>or 1+0, there is one caveat you have to consider...
>
>the no.of arms or disks available to your application(and Oracle) for
>servicing read/write requests is cut to half with RAID 0+1. These days,
>you can get 300 GB disks and a Terabyte worth of data can sit in about
>6-8 disks! If you are running a r/w intensive application, no amount of
>IO balancing can help..because you don't have anything to balance! just
>3-4 disks will be servicing all the needs of the application, don't
>even bother about the no.of users.
>
>the key is IOps/seconds. What the peak and sustainable IO rate that the
>SAN box or disk system provide? what are the needs of your application?
>how does it access the data? predominantly reads or writes, etc.?
>
>Whereas with RAID 5, you can have 70% of your disks servicing your
>read/write requests. I have seen many IO intensive application use RAID
>5 successfully and with S.A.M.E technology. You have to get it right,
>then it works for you. There is, of course, a well documented slowness
>in write thoroughout when you compare it with RAID 0+1.
>
>For your 2.2 TB SAN Box, you would surely have some cache. I would
>suspect 6-8GB may be available. So, that also needs to be considered.
>
>Good luck!
>
>cheers
>anand
This is truely a sad post. Yet another advocate of 'throwing iron at the problem' ie symptom fighting, instead of curing the problem.
If you have a R/W intensive application, and the customer doesn't want to resolve the problem (because the app has been developed by an incompetent 3rd party vendor), NOTHING can help you, only PRAYER, there is some justice in Heaven (and the customer will suffer so badly, he can't deny the facts anymore)
-- Sybrand Bakker, Senior Oracle DBAReceived on Fri Dec 16 2005 - 02:19:42 CST
![]() |
![]() |