Re: Is Raid 5 really that bad for Oracle?

From: Daniel Morgan <damorgan_at_x.washington.edu>
Date: Tue, 03 Aug 2004 07:21:23 -0700
Message-ID: <1091542918.68711@yasure>

Connor McDonald wrote:

> Niall Litchfield wrote:
>

>>"Cary Millsap" <cary.millsap .at. hotsos .dot. com> wrote in message news:<10gu902j8cmcs7b_at_corp.supernews.com>...
>>
>>>2. But mirroring is expensive per byte of storage. Hence RAID levels 2, 3,
>>>4, 5, and 6 were proposed. Their design goals were to lessen the
>>>expense-per-byte of storage of RAID level 1 (mirroring). For example, with
>>>G=5 RAID level 5, the price of resilience per byte of storage is 5/4 of a
>>>4-disk array instead of 8/4.
>>
>>I meant to make a similar point originally, but didn't. Namely that
>>RAID5 tends to make sense from a COST/GB point of view and SAME tends
>>to make sense from a COST/IO point of view. Often these two views are
>>where the disagreement lies. If my sysadmins ask me how much storage I
>>want, they do not want to be told (say) 100gb and 2500 IO/sec. They
>>just want the former figure.
>>
>>Niall

>
>
> True story...
>
> Company X moving from JBOD to SAN. JBOD totals around 800G so the SAN
> is populated to 800G as well...Of course, in the JBOD, each 18G disk
> ends up being around 17.5G available to the OS. In the SAN, each 36G
> disk ends up being around 30G to the OS
>
> Result:
> - They've halved the number of spindles
> - They've not enough disk space
> but hey...at least we've got a SAN
>
> :-)

Good point. But is the database really writing to those spindles or is it, more likely, writing to a very large RAM cache rendering the loss irrelevant?

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace 'x' with 'u' to respond)

Received on Tue Aug 03 2004 - 09:21:23 CDT