<?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: Network virtualisation</title>
	<atom:link href="http://veejoe.net/blog/2010/02/network-virtualisation/feed/" rel="self" type="application/rss+xml" />
	<link>http://veejoe.net/blog/2010/02/network-virtualisation/</link>
	<description>Vic's Blog</description>
	<lastBuildDate>Fri, 18 May 2012 15:44:00 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.3.2</generator>
	<item>
		<title>By: Edouard Bourguignon</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-465</link>
		<dc:creator>Edouard Bourguignon</dc:creator>
		<pubDate>Thu, 19 May 2011 15:22:00 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-465</guid>
		<description> Thank you, I will try to get this issue of Admin Magazine</description>
		<content:encoded><![CDATA[<p> Thank you, I will try to get this issue of Admin Magazine</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Vic Cross</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-426</link>
		<dc:creator>Vic Cross</dc:creator>
		<pubDate>Tue, 10 May 2011 14:32:00 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-426</guid>
		<description>Thanks for your comment...  I have to admit it took me a while to understand again what the problem was with the way I was trying to work it.

The problem was that I wanted different virtual machines to be configured differently -- some to receive untagged frames for particular VLANs, and others to receive all the tagged frames straight from the switch.  I can&#039;t recall if I tried your method or not, but what I found was that the physical interface no longer saw any packets once a VLAN was configured against it, which meant the &quot;VLAN-aware bridge&quot; didn&#039;t see any traffic.  It didn&#039;t appear that a bridge was sufficient isolation of that (the second paragraph in my original post implies that I tried something like what you described but it didn&#039;t work for me).  Perhaps the choice of network card makes a difference too, changing where the VLAN processing happens...?

Like I said though, I can&#039;t recall if I tried a bridge behind a bridge as you&#039;ve shown -- perhaps I should give it a go!</description>
		<content:encoded><![CDATA[<p>Thanks for your comment&#8230;  I have to admit it took me a while to understand again what the problem was with the way I was trying to work it.</p>
<p>The problem was that I wanted different virtual machines to be configured differently &#8212; some to receive untagged frames for particular VLANs, and others to receive all the tagged frames straight from the switch.  I can&#8217;t recall if I tried your method or not, but what I found was that the physical interface no longer saw any packets once a VLAN was configured against it, which meant the &#8220;VLAN-aware bridge&#8221; didn&#8217;t see any traffic.  It didn&#8217;t appear that a bridge was sufficient isolation of that (the second paragraph in my original post implies that I tried something like what you described but it didn&#8217;t work for me).  Perhaps the choice of network card makes a difference too, changing where the VLAN processing happens&#8230;?</p>
<p>Like I said though, I can&#8217;t recall if I tried a bridge behind a bridge as you&#8217;ve shown &#8212; perhaps I should give it a go!</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: veejoe</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-423</link>
		<dc:creator>veejoe</dc:creator>
		<pubDate>Tue, 10 May 2011 13:51:00 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-423</guid>
		<description>Yes, I was doing something very similar to what you describe.  I was having trouble trying to use the brcompat functionality to make Open vSwitch fit into the Ubuntu network startup method, but I think I&#039;ve now learned that I need to throw away the normal network startup and let Open vSwitch take care of everything (unless Ubuntu Server has native support for Open vSwitch nowadays...).

There is a nice looking writeup of Open vSwitch in Issue 03 of ADMIN Magazine; the author discusses the use of fake bridges to achieve PVID-like function.</description>
		<content:encoded><![CDATA[<p>Yes, I was doing something very similar to what you describe.  I was having trouble trying to use the brcompat functionality to make Open vSwitch fit into the Ubuntu network startup method, but I think I&#8217;ve now learned that I need to throw away the normal network startup and let Open vSwitch take care of everything (unless Ubuntu Server has native support for Open vSwitch nowadays&#8230;).</p>
<p>There is a nice looking writeup of Open vSwitch in Issue 03 of ADMIN Magazine; the author discusses the use of fake bridges to achieve PVID-like function.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Edouard Bourguignon</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-418</link>
		<dc:creator>Edouard Bourguignon</dc:creator>
		<pubDate>Thu, 28 Apr 2011 12:54:00 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-418</guid>
		<description>Hi,

Do you know if openvswitch can tag untagged traffic? Like PVID functionnality on physical switch. Because our VM boots on PXE, and PXE doesn&#039;t support VLAN. So we have to tag our vswitch port to 0 to allow untagged traffic from the VM to pass thru. Is it possible that fake bridge could handle that traffic to a tagged port?</description>
		<content:encoded><![CDATA[<p>Hi,</p>
<p>Do you know if openvswitch can tag untagged traffic? Like PVID functionnality on physical switch. Because our VM boots on PXE, and PXE doesn&#8217;t support VLAN. So we have to tag our vswitch port to 0 to allow untagged traffic from the VM to pass thru. Is it possible that fake bridge could handle that traffic to a tagged port?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: telmich</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-302</link>
		<dc:creator>telmich</dc:creator>
		<pubDate>Sat, 22 Jan 2011 08:58:00 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-302</guid>
		<description>I&#039;m also a bit confused, it looks like this should work directly with one additional interface: Assume eth0 is the physical interface of the vm-host which receives and sends tagged frames (8021q). Assume further that there&#039;s a bridge named brtag, into which eth0 is bound. Then you add a vlan enabled device called vlan123 on brtag, that adds the vlan tag 123. Then you create a bridge named &quot;brvlan123&quot; and add the interface vlan123 to it. Guests which shall set tags themselves are added to brtag, guests which should not setup tags are bound to brvlan123.nnMaybe I&#039;m misunderstanding your post, but I do not see a need for binding one interface to multiple bridges.</description>
		<content:encoded><![CDATA[<p>I&#8217;m also a bit confused, it looks like this should work directly with one additional interface: Assume eth0 is the physical interface of the vm-host which receives and sends tagged frames (8021q). Assume further that there&#8217;s a bridge named brtag, into which eth0 is bound. Then you add a vlan enabled device called vlan123 on brtag, that adds the vlan tag 123. Then you create a bridge named &#8220;brvlan123&#8243; and add the interface vlan123 to it. Guests which shall set tags themselves are added to brtag, guests which should not setup tags are bound to brvlan123.nnMaybe I&#8217;m misunderstanding your post, but I do not see a need for binding one interface to multiple bridges.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Vic</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-270</link>
		<dc:creator>Vic</dc:creator>
		<pubDate>Sat, 28 Aug 2010 13:29:08 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-270</guid>
		<description>Hi Guido,

I have used that method, when running VMware Server in the past.  Trouble is, it doesn&#039;t work when you have a mixture of VLAN-aware and VLAN-unaware guests that you want to share an adapter: as soon as the 8021q module is bound to an interface (when a VLAN is configured) the &quot;non-VLAN&quot; or &quot;raw&quot; interface no longer receives all the packets.  In my VMware Server days, I had to allocate an additional NIC with no VLAN membership for VLAN-aware guests.

I thought that OpenVswitch would be a way to provide more flexibility, and allow me greater control to do interesting things with my available network hardware (especially since I was moving to a new machine with reduced capacity for NIC installation).  Unfortunately I had a lot of trouble integrating OpenVswitch into the boot process of Ubuntu Lucid -- every bootup required manual network reconfiguration.  I have not looked at it for a while though, hopefully there&#039;s better support in distros now.

Thanks for your comment!

Vic
</description>
		<content:encoded><![CDATA[<p>Hi Guido,</p>
<p>I have used that method, when running VMware Server in the past.  Trouble is, it doesn&#8217;t work when you have a mixture of VLAN-aware and VLAN-unaware guests that you want to share an adapter: as soon as the 8021q module is bound to an interface (when a VLAN is configured) the &#8220;non-VLAN&#8221; or &#8220;raw&#8221; interface no longer receives all the packets.  In my VMware Server days, I had to allocate an additional NIC with no VLAN membership for VLAN-aware guests.</p>
<p>I thought that OpenVswitch would be a way to provide more flexibility, and allow me greater control to do interesting things with my available network hardware (especially since I was moving to a new machine with reduced capacity for NIC installation).  Unfortunately I had a lot of trouble integrating OpenVswitch into the boot process of Ubuntu Lucid &#8212; every bootup required manual network reconfiguration.  I have not looked at it for a while though, hopefully there&#8217;s better support in distros now.</p>
<p>Thanks for your comment!</p>
<p>Vic</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Guido Trotter</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-178</link>
		<dc:creator>Guido Trotter</dc:creator>
		<pubDate>Fri, 20 Aug 2010 12:45:32 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-178</guid>
		<description>What&#039;s the advantage of using Openvswitch over simply configuring 802.1Q interfaces on linux (for example with: ip link add link eth0 name eth0.3 type vlan id 3) and then connecting those (in this example eth0.3) to a bridge:
brctl addbr br3
brctl addbr addif br3 eth0.3

Sounds like you&#039;d get the same effect without patching your kernel.

Thanks,

Guido</description>
		<content:encoded><![CDATA[<p>What&#8217;s the advantage of using Openvswitch over simply configuring 802.1Q interfaces on linux (for example with: ip link add link eth0 name eth0.3 type vlan id 3) and then connecting those (in this example eth0.3) to a bridge:<br />
brctl addbr br3<br />
brctl addbr addif br3 eth0.3</p>
<p>Sounds like you&#8217;d get the same effect without patching your kernel.</p>
<p>Thanks,</p>
<p>Guido</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: deshantm</title>
		<link>http://veejoe.net/blog/2010/02/network-virtualisation/comment-page-1/#comment-83</link>
		<dc:creator>deshantm</dc:creator>
		<pubDate>Thu, 04 Mar 2010 00:38:01 +0000</pubDate>
		<guid isPermaLink="false">http://veejoe.net/blog/?p=5494#comment-83</guid>
		<description>Thanks for this post. I have also been trying open vSwitch with KVM lately, for a different project. I was trying to figure if libvirt could use a custom script, but it seems they have disabled that functionality [1]. I was then looking at the libvirt code, and it didn&#039;t look that hard to add, but it would be a hack. Hacking the default /etc/qemu-ifup makes more sense. I will try that. Thanks again for the idea.&lt;br&gt;&lt;br&gt;&lt;br&gt;[1] &lt;a href=&quot;http://www.mail-archive.com/libvir-list%40redhat.com/msg18133.html&quot; rel=&quot;nofollow&quot;&gt;http://www.mail-archive.com/libvir-list@redhat....&lt;/a&gt;</description>
		<content:encoded><![CDATA[<p>Thanks for this post. I have also been trying open vSwitch with KVM lately, for a different project. I was trying to figure if libvirt could use a custom script, but it seems they have disabled that functionality [1]. I was then looking at the libvirt code, and it didn&#39;t look that hard to add, but it would be a hack. Hacking the default /etc/qemu-ifup makes more sense. I will try that. Thanks again for the idea.</p>
<p>[1] <a href="http://www.mail-archive.com/libvir-list%40redhat.com/msg18133.html" rel="nofollow"></a><a href="http://www.mail-archive.com/libvir-list@redhat" rel="nofollow">http://www.mail-archive.com/libvir-list@redhat</a>&#8230;.</p>
]]></content:encoded>
	</item>
</channel>
</rss>

