<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: 3010: A MultiStore Primer</title>
	<atom:link href="http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/feed/" rel="self" type="application/rss+xml" />
	<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/</link>
	<description>The weblog of an IT pro specializing in virtualization, storage, and servers</description>
	<lastBuildDate>Wed, 08 Feb 2012 17:13:47 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=</generator>
	<item>
		<title>By: slowe</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44442</link>
		<dc:creator>slowe</dc:creator>
		<pubDate>Tue, 12 May 2009 11:48:03 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44442</guid>
		<description>Scott G,

Don&#039;t worry about catching it late--after all, I posted it many, many months late.  :-)</description>
		<content:encoded><![CDATA[<p>Scott G,</p>
<p>Don&#8217;t worry about catching it late&#8211;after all, I posted it many, many months late.  <img src='http://blog.scottlowe.org/wp-includes/images/smilies/icon_smile.gif' alt=':-)' class='wp-smiley' /> </p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Scott Gelb</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44438</link>
		<dc:creator>Scott Gelb</dc:creator>
		<pubDate>Tue, 12 May 2009 05:25:26 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44438</guid>
		<description>Scott,

Cool you posted this...thank you.  I only caught it a month late :)  Roger posted our presentation to the partner portal on the NetApp Insight page, and he also posted a lighter version of it on PartnerCenter for employees and partners to reference.  Also, I can send you or ftp the 11 minute avi video from the presentation.  Nothing proprietary and running from my laptop showing both vfiler migrate and vfiler dr.

MultiStore does support FTP.  Status of a vFiler below that shows all protocols for a vFiler.  All are allowed by default (we have a request in for future ONTAP to disallow by default then have the administrator allow protocols they want).

   Allowed: proto=rsh
   Allowed: proto=ssh
   Allowed: proto=nfs
   Allowed: proto=cifs
   Allowed: proto=iscsi
   Allowed: proto=ftp
   Allowed: proto=http

One other note... vfiler dr doesn&#039;t change the volume option &quot;fs_size_fixed&quot; from on to off with &quot;vfiler dr activate&quot;.  So, if the DR site is running for a long period of time before resync back to the repaired source, and you need to grow a volume, you&#039;ll have to turn this option off to utilize additional space added to a volume.

To elaborate on qtrees, we used to use qtrees more with vFilers before ONTAP 7G (to better utilize legacy traditional volumes).  With FlexVols we have the same utilization benefit we had with qtrees on tradvols.  When using FlexVols instead of qtrees, vFiler dr/migrate use volume mirror which give the advantage of keeping dedup over the wire (not rehydrating data like qsm), and also if we leverage snapmover (vfiler migrate -m nocopy) then we need the vFiler to own all FlexVols in the aggregates used for the vFiler.  Also, if an nfs mount is on a qtree and the parent volume of the qtree is not owned by the vFiler, a vFiler migrate will migrate the qtree with qsm and the target qtree will not be accessible without a client remount (volume fsid isn&#039;t changed).  However, if the parent FlexVol is owned by the vFiler, then the qtree mount is accessible since VSM is used and the fsid and file handles are moved to the target FlexVol.  Nothing against qtrees though...I still use them a lot but don&#039;t assign individual qtrees to vFilers.

We do some cool things with vFilers for DMZ readonly data sharing...use a FlexClone or a SnapMirror of data in one vFiler to another.  You can use the loopback adapter but running the mirror from vfiler0, then you don&#039;t need any network connection between the vfilers and it&#039;s still secure but can have a copy of any data you want...

For management, definitely a lot of cli.  If customers use the cli already they are comfortable.  if they use FilerView, then they have some learning curve.  API has support for vFilers...so hopefully the new NSM will support vfilers in a future release.

Another note...it&#039;s worth a day (often less than a day) of PS to get MultiStore setup properly....it&#039;s not rocket science, but those who work with it a lot can do in minutes what might take others hours so it saves cost to pay for a day or half a day for setup.  For example, the /etc/rc file isn&#039;t always updated correctly, but is easy to edit after vfiler creation/configuration.  For example, if you use an ip alias or vlan, sometimes the commands are put above the vif creation.  For example, the alias to vif10 is made before vif10 is created, so on boot the alias will fail.  Also, it is important to add the route for the vfiler if needed &quot;vfiler run route add default&quot; in the rc file.  Also, a lot of customers implement MultiStore after the system is already running...so we move the resources in vFiler0 to another vFiler that can be used for DR...in this case we often don&#039;t want to initialize mirrors (vfiler dr configure will initialize mirrors even if initialized), but instead we just manually mirror the vfiler root volume created on the source, then create the dr vfiler manually, then resync it...saves the pain of re-init of a mirror.

Another cool thing... ONTAP 7.3.1 added one of the items on our wish list for the secure dr feature using ssl, so we don&#039;t have to have rsh enabled for the controller to controller communication.  &quot;vfiler dr configure [-c secure]&quot;  7.3.1 also put sis (dedup) commands in the vFiler context now too, so it&#039;s configurable from within the vFiler itself now.

MultiStore Rocks :)

-Scott Gelb</description>
		<content:encoded><![CDATA[<p>Scott,</p>
<p>Cool you posted this&#8230;thank you.  I only caught it a month late <img src='http://blog.scottlowe.org/wp-includes/images/smilies/icon_smile.gif' alt=':)' class='wp-smiley' />   Roger posted our presentation to the partner portal on the NetApp Insight page, and he also posted a lighter version of it on PartnerCenter for employees and partners to reference.  Also, I can send you or ftp the 11 minute avi video from the presentation.  Nothing proprietary and running from my laptop showing both vfiler migrate and vfiler dr.</p>
<p>MultiStore does support FTP.  Status of a vFiler below that shows all protocols for a vFiler.  All are allowed by default (we have a request in for future ONTAP to disallow by default then have the administrator allow protocols they want).</p>
<p>   Allowed: proto=rsh<br />
   Allowed: proto=ssh<br />
   Allowed: proto=nfs<br />
   Allowed: proto=cifs<br />
   Allowed: proto=iscsi<br />
   Allowed: proto=ftp<br />
   Allowed: proto=http</p>
<p>One other note&#8230; vfiler dr doesn&#8217;t change the volume option &#8220;fs_size_fixed&#8221; from on to off with &#8220;vfiler dr activate&#8221;.  So, if the DR site is running for a long period of time before resync back to the repaired source, and you need to grow a volume, you&#8217;ll have to turn this option off to utilize additional space added to a volume.</p>
<p>To elaborate on qtrees, we used to use qtrees more with vFilers before ONTAP 7G (to better utilize legacy traditional volumes).  With FlexVols we have the same utilization benefit we had with qtrees on tradvols.  When using FlexVols instead of qtrees, vFiler dr/migrate use volume mirror which give the advantage of keeping dedup over the wire (not rehydrating data like qsm), and also if we leverage snapmover (vfiler migrate -m nocopy) then we need the vFiler to own all FlexVols in the aggregates used for the vFiler.  Also, if an nfs mount is on a qtree and the parent volume of the qtree is not owned by the vFiler, a vFiler migrate will migrate the qtree with qsm and the target qtree will not be accessible without a client remount (volume fsid isn&#8217;t changed).  However, if the parent FlexVol is owned by the vFiler, then the qtree mount is accessible since VSM is used and the fsid and file handles are moved to the target FlexVol.  Nothing against qtrees though&#8230;I still use them a lot but don&#8217;t assign individual qtrees to vFilers.</p>
<p>We do some cool things with vFilers for DMZ readonly data sharing&#8230;use a FlexClone or a SnapMirror of data in one vFiler to another.  You can use the loopback adapter but running the mirror from vfiler0, then you don&#8217;t need any network connection between the vfilers and it&#8217;s still secure but can have a copy of any data you want&#8230;</p>
<p>For management, definitely a lot of cli.  If customers use the cli already they are comfortable.  if they use FilerView, then they have some learning curve.  API has support for vFilers&#8230;so hopefully the new NSM will support vfilers in a future release.</p>
<p>Another note&#8230;it&#8217;s worth a day (often less than a day) of PS to get MultiStore setup properly&#8230;.it&#8217;s not rocket science, but those who work with it a lot can do in minutes what might take others hours so it saves cost to pay for a day or half a day for setup.  For example, the /etc/rc file isn&#8217;t always updated correctly, but is easy to edit after vfiler creation/configuration.  For example, if you use an ip alias or vlan, sometimes the commands are put above the vif creation.  For example, the alias to vif10 is made before vif10 is created, so on boot the alias will fail.  Also, it is important to add the route for the vfiler if needed &#8220;vfiler run route add default&#8221; in the rc file.  Also, a lot of customers implement MultiStore after the system is already running&#8230;so we move the resources in vFiler0 to another vFiler that can be used for DR&#8230;in this case we often don&#8217;t want to initialize mirrors (vfiler dr configure will initialize mirrors even if initialized), but instead we just manually mirror the vfiler root volume created on the source, then create the dr vfiler manually, then resync it&#8230;saves the pain of re-init of a mirror.</p>
<p>Another cool thing&#8230; ONTAP 7.3.1 added one of the items on our wish list for the secure dr feature using ssl, so we don&#8217;t have to have rsh enabled for the controller to controller communication.  &#8220;vfiler dr configure [-c secure]&#8221;  7.3.1 also put sis (dedup) commands in the vFiler context now too, so it&#8217;s configurable from within the vFiler itself now.</p>
<p>MultiStore Rocks <img src='http://blog.scottlowe.org/wp-includes/images/smilies/icon_smile.gif' alt=':)' class='wp-smiley' /> </p>
<p>-Scott Gelb</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Vaughn</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44260</link>
		<dc:creator>Vaughn</dc:creator>
		<pubDate>Sat, 18 Apr 2009 02:34:33 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44260</guid>
		<description>I expanded this discussion by discussing intergrating vFilers with VMware SRM.  Check it out if interested.

http://blogs.netapp.com/virtualstorageguy/2009/04/site-recovery-manager-its-not-just-for-vms.html

Cheers,
Vaughn</description>
		<content:encoded><![CDATA[<p>I expanded this discussion by discussing intergrating vFilers with VMware SRM.  Check it out if interested.</p>
<p><a href="http://blogs.netapp.com/virtualstorageguy/2009/04/site-recovery-manager-its-not-just-for-vms.html" rel="nofollow">http://blogs.netapp.com/virtualstorageguy/2009/04/site-recovery-manager-its-not-just-for-vms.html</a></p>
<p>Cheers,<br />
Vaughn</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Dan C</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44230</link>
		<dc:creator>Dan C</dc:creator>
		<pubDate>Tue, 14 Apr 2009 13:54:27 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44230</guid>
		<description>William&gt; Sure, it can.</description>
		<content:encoded><![CDATA[<p>William&gt; Sure, it can.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: JeffA</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44198</link>
		<dc:creator>JeffA</dc:creator>
		<pubDate>Fri, 10 Apr 2009 00:58:28 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44198</guid>
		<description>I totally disagree Tim.  Qtrees are useless in only 98% of environments at most!

Actually, one real benefit to Qtrees is SnapMirror.  Back when I used to deploy systems instead of just sell them, I would make the root of the flexvol empty and put in a single Qtree.  Then I would build Qtree SnapMirror (QSM) relationships instead of Volume SnapMirrors (VSM).  The benefit is that the volume sizes could be different and I could resize the volumes more easily than with VSM.  Sure, there is a performance penalty, but in most of the environments I worked in the perf penalty was never felt and the easy flexibility was good to have in your pocket.</description>
		<content:encoded><![CDATA[<p>I totally disagree Tim.  Qtrees are useless in only 98% of environments at most!</p>
<p>Actually, one real benefit to Qtrees is SnapMirror.  Back when I used to deploy systems instead of just sell them, I would make the root of the flexvol empty and put in a single Qtree.  Then I would build Qtree SnapMirror (QSM) relationships instead of Volume SnapMirrors (VSM).  The benefit is that the volume sizes could be different and I could resize the volumes more easily than with VSM.  Sure, there is a performance penalty, but in most of the environments I worked in the perf penalty was never felt and the easy flexibility was good to have in your pocket.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Fletch</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44197</link>
		<dc:creator>Fletch</dc:creator>
		<pubDate>Thu, 09 Apr 2009 23:53:01 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44197</guid>
		<description>Scott, great write up - we&#039;ve been running our VI off Netapp NFS vFilers for close to 2 years now.  We turned on dedup with the 7.3 upgrade - was great to see the instant (well overnight anyway) savings 5Tb!
We use vfiler dr between 2 campus area clusters and also to a 3rd dr site (chained snapmirrors) for our website content.
I can&#039;t wait until Netapp updates their SRA to support VMWare SRM to automate DR...as it stands the vfiler dr failover and back steps are fairly quick...</description>
		<content:encoded><![CDATA[<p>Scott, great write up &#8211; we&#8217;ve been running our VI off Netapp NFS vFilers for close to 2 years now.  We turned on dedup with the 7.3 upgrade &#8211; was great to see the instant (well overnight anyway) savings 5Tb!<br />
We use vfiler dr between 2 campus area clusters and also to a 3rd dr site (chained snapmirrors) for our website content.<br />
I can&#8217;t wait until Netapp updates their SRA to support VMWare SRM to automate DR&#8230;as it stands the vfiler dr failover and back steps are fairly quick&#8230;</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: TimC</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44196</link>
		<dc:creator>TimC</dc:creator>
		<pubDate>Thu, 09 Apr 2009 20:37:11 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44196</guid>
		<description>@JeffA: That it was.  But with flexvol&#039;s, using qtree&#039;s is significantly less interesting.  Qtree&#039;s in general I would argue have basically become a moot point in 99% of the environments out there with the advent of flexvol&#039;s.</description>
		<content:encoded><![CDATA[<p>@JeffA: That it was.  But with flexvol&#8217;s, using qtree&#8217;s is significantly less interesting.  Qtree&#8217;s in general I would argue have basically become a moot point in 99% of the environments out there with the advent of flexvol&#8217;s.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: JeffA</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44193</link>
		<dc:creator>JeffA</dc:creator>
		<pubDate>Thu, 09 Apr 2009 05:16:38 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44193</guid>
		<description>I attended this session at Insight last year and presented a very similar session to the Charlotte Tech OnTap Live group last fall.  

One small thing to add is that you can use qtrees instead of volumes for the vfiler root.  That was a big deal back in the Data ONTAP 6.X days.

Good stuff as usual, Scott.</description>
		<content:encoded><![CDATA[<p>I attended this session at Insight last year and presented a very similar session to the Charlotte Tech OnTap Live group last fall.  </p>
<p>One small thing to add is that you can use qtrees instead of volumes for the vfiler root.  That was a big deal back in the Data ONTAP 6.X days.</p>
<p>Good stuff as usual, Scott.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: William</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44191</link>
		<dc:creator>William</dc:creator>
		<pubDate>Thu, 09 Apr 2009 04:45:06 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44191</guid>
		<description>Can the vFiler be managed by NetApp OnTap SDK? I recently came across this blog awhile back and it had mentioned Powershell and NetApp: http://get-admin.com/blog/?p=545</description>
		<content:encoded><![CDATA[<p>Can the vFiler be managed by NetApp OnTap SDK? I recently came across this blog awhile back and it had mentioned Powershell and NetApp: <a href="http://get-admin.com/blog/?p=545" rel="nofollow">http://get-admin.com/blog/?p=545</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ausmith1</title>
		<link>http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/comment-page-1/#comment-44190</link>
		<dc:creator>Ausmith1</dc:creator>
		<pubDate>Thu, 09 Apr 2009 02:19:09 +0000</pubDate>
		<guid isPermaLink="false">http://blog.scottlowe.org/2009/04/08/3010-a-multistore-primer/#comment-44190</guid>
		<description>You can&#039;t directly ssh to a specific vfiler context but if you you use a scriptable ssh tool (such as SecureCRT) you can have it automatically log into the desired context, see example below:

usbvtrisicr200&gt; vfiler context usbvtcommonfiles
usbvtcommonfiles@usbvtrisicr200&gt; Wed Apr  8 22:18:24 EDT [usbvtcommonfiles@cmds.vfiler.console.switch:notice]: Console context was switched to a vFiler(tm) unit usbvtcommonfiles.

usbvtcommonfiles@usbvtrisicr200&gt;</description>
		<content:encoded><![CDATA[<p>You can&#8217;t directly ssh to a specific vfiler context but if you you use a scriptable ssh tool (such as SecureCRT) you can have it automatically log into the desired context, see example below:</p>
<p>usbvtrisicr200&gt; vfiler context usbvtcommonfiles<br />
usbvtcommonfiles@usbvtrisicr200&gt; Wed Apr  8 22:18:24 EDT [usbvtcommonfiles@cmds.vfiler.console.switch:notice]: Console context was switched to a vFiler(tm) unit usbvtcommonfiles.</p>
<p>usbvtcommonfiles@usbvtrisicr200&gt;</p>
]]></content:encoded>
	</item>
</channel>
</rss>

