<div dir="ltr">Is there plans to put 1.11 into the Ubuntu Cloud Archive for Havana?</div><div class="gmail_extra"><br><br><div class="gmail_quote">On Sat, Oct 26, 2013 at 10:29 AM, George Shuklin <span dir="ltr"><<a href="mailto:george.shuklin@gmail.com" target="_blank">george.shuklin@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000">
Multithreading is not an issue. <br>
<br>
Main feature in ovs 1.11 which make it 'wild internet ready' is
'megaflow'. <br>
<br>
Problem is that in old mode kernel module send every new flow header
to inspection to userspace daemon. And daemon send rule to kernel
(ovs-dpctl dump-flows). Prior 1.11 that rule (internal rule, not
'openflow') could not contain masks. So in --flood mode it simply
cause every new packet going to slow path overloading daemon.<br>
<br>
In OVS 1.11 now internal rules can contain masks (megaflow it is)
and now different port numbers, source/dest IPs is no longer an
issue.<div><div class="h5"><br>
<br>
<br>
<div>On 26.10.2013 12:45, Édouard Thuleau
wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">We made some tests to know the limit of
simultaneous established connections a VM can support and we met
some pb with 1.11 release (crashes...).
<div><br>
<div>So we try the new release 2.0.0 [1] (on the advice of KVM
list [2] ) which had the support for the multiple threads
for ovs-vswitchd process. And now, we no longer have
limitations from OVS.</div>
<div><br>
</div>
<div>Ubuntu ppa packages OVS 1.11.0 and 2.0.0 are available
here [3] thanks to Sahid.</div>
</div>
<div><br>
</div>
<div>[1] <a href="http://git.openvswitch.org/cgi-bin/gitweb.cgi?p=openvswitch;a=blob;f=NEWS;h=e0fafd08f52cc3a753cef6554251b9288cd3e6e3;hb=refs/heads/branch-2.0" target="_blank">http://git.openvswitch.org/cgi-bin/gitweb.cgi?p=openvswitch;a=blob;f=NEWS;h=e0fafd08f52cc3a753cef6554251b9288cd3e6e3;hb=refs/heads/branch-2.0</a></div>
<div>[2] <a href="http://www.mail-archive.com/kvm@vger.kernel.org/msg96823.html" target="_blank">http://www.mail-archive.com/kvm@vger.kernel.org/msg96823.html</a></div>
<div>[3] <a href="https://launchpad.net/%7Esahid-ferdjaoui/+archive/openvswitch" target="_blank">https://launchpad.net/~sahid-ferdjaoui/+archive/openvswitch</a></div>
<div><br>
</div>
<div>Regards,</div>
<div>Édouard.</div>
<div><br>
</div>
</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">On Sat, Oct 26, 2013 at 2:26 AM, George
Shuklin <span dir="ltr"><<a href="mailto:george.shuklin@gmail.com" target="_blank">george.shuklin@gmail.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<p dir="ltr">I my tests any ovs below 1.11 suck hardly. You
can see it by simple hping3 --flood vm.ip</p>
<p dir="ltr">All older ovs simply can't keep pace over
10-15mb/sec. Ovs 1.11easely operates at 8-10 gb/s.</p>
<p dir="ltr">I don't know about is stability, but putting
anything except it with naked butt toward hostile internet
is really not wize. It's like invitation for script
kiddie: DoS me please from your cellular internet.</p>
<br>
_______________________________________________<br>
OpenStack-operators mailing list<br>
<a href="mailto:OpenStack-operators@lists.openstack.org" target="_blank">OpenStack-operators@lists.openstack.org</a><br>
<a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators</a><br>
<br>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</div></div></div>
<br>_______________________________________________<br>
OpenStack-operators mailing list<br>
<a href="mailto:OpenStack-operators@lists.openstack.org">OpenStack-operators@lists.openstack.org</a><br>
<a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators</a><br>
<br></blockquote></div><br></div>