<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/1.5.2" -->
<rss version="2.0" 
	xmlns:content="http://purl.org/rss/1.0/modules/content/">
<channel>
	<title>Comments on: RAID Stripe width</title>
	<link>http://storageadvisors.adaptec.com/2007/09/07/raid-stripe-width/</link>
	<description>Storage Solutions for Real World IT Professionals</description>
	<pubDate>Sat, 05 Jul 2008 07:56:57 +0000</pubDate>
	<generator>http://wordpress.org/?v=1.5.2</generator>

	<item>
 		<title>Comment on RAID Stripe width by: maobo</title>
		<link>http://storageadvisors.adaptec.com/2007/09/07/raid-stripe-width/#comment-80977</link>
		<pubDate>Tue, 16 Oct 2007 00:17:30 +0000</pubDate>
		<guid>http://storageadvisors.adaptec.com/2007/09/07/raid-stripe-width/#comment-80977</guid>
					<description>I want to ask a question about the read/write cache size ratio in the real storage controller. As we know the write cache is usually a battery-backed RAM in storage controller. But for battery life limit, its size are restrict to a small size. What the really ratio in the common storage controller? Thank you very much.</description>
		<content:encoded><![CDATA[	<p>I want to ask a question about the read/write cache size ratio in the real storage controller. As we know the write cache is usually a battery-backed RAM in storage controller. But for battery life limit, its size are restrict to a small size. What the really ratio in the common storage controller? Thank you very much.
</p>
]]></content:encoded>
				</item>
	<item>
 		<title>Comment on RAID Stripe width by: Alex</title>
		<link>http://storageadvisors.adaptec.com/2007/09/07/raid-stripe-width/#comment-76611</link>
		<pubDate>Tue, 25 Sep 2007 08:24:19 +0000</pubDate>
		<guid>http://storageadvisors.adaptec.com/2007/09/07/raid-stripe-width/#comment-76611</guid>
					<description>Thanks for the info. 

I have seen a post on the web that says RAID-5 write performance degrades with more disks, because:

&quot;writing to a RAID-5 random location requires 2 writes (to the block and it's parity) plus N-2 reads (so you have something to compute the parity with).  As the number of drives in the RAID-5 set gets larger, the write performance gets worse. &quot;

I assume the N-2 comment is wrong (where n is the stripe size)? A RAID-5 write will always take 4 operations regardless of stripe size (in a non-degraded state, of course)

One further point: Will an 8-disk RAID 5 or 2 x 4-disk RAID 5 arrays perform similarly for writes, as long as the writes are distributed evenly between the two arrays?

cheers

alex</description>
		<content:encoded><![CDATA[	<p>Thanks for the info. </p>
	<p>I have seen a post on the web that says RAID-5 write performance degrades with more disks, because:</p>
	<p>&#8220;writing to a RAID-5 random location requires 2 writes (to the block and it&#8217;s parity) plus N-2 reads (so you have something to compute the parity with).  As the number of drives in the RAID-5 set gets larger, the write performance gets worse. &#8221;</p>
	<p>I assume the N-2 comment is wrong (where n is the stripe size)? A RAID-5 write will always take 4 operations regardless of stripe size (in a non-degraded state, of course)</p>
	<p>One further point: Will an 8-disk RAID 5 or 2 x 4-disk RAID 5 arrays perform similarly for writes, as long as the writes are distributed evenly between the two arrays?</p>
	<p>cheers</p>
	<p>alex
</p>
]]></content:encoded>
				</item>
</channel>
</rss>
