[Openstack] Changing quantum/neutron OVS Plugin

Xin Zhao xzhao at bnl.gov
Fri Sep 13 13:48:20 UTC 2013


I have a similar question. We are considering to upgrade to grizzly 
using OVS/vlan model, if I understand the doc correctly,
all the external traffic and internal intra-virtual network traffic go 
through the network host, which makes the network host a
single point of failure, and high loaded. So my question is, how people 
deal with this bottleneck in network node? Is it possible
to deploy multiple network nodes, or using other plugin, like OpenFlow, 
is the solution ?


On 9/11/2013 3:29 AM, Geraint Jones wrote:
> Hi
> We have deployed into production a cluster using OVS.
> We are using "tenant_network_type = gre"
> We have two controller nodes (these also run ceph for storage) each 
> have 2 nic's bonded using VLAN's for the different traffic (Storage on 
> one VLAN, VM Mgmt on another etc)
> We have a single network node (this is an oldish box with a quad 
> core X3220) this has two nic's one is a public facing nic, the other 
> has two VLANs on it (VM Mgmt and VM Floating)
> We have eight compute nodes these are dual nic's bonded with the same 
> VLANs as the controller and network nodes.
> We are facing a few issues.
> 1.) The network node has nic's that cannot exceed 1500 byte MTU -- so 
> all vms need to have a small MTU to avoid fragmentation -- this is a 
> pain in the ass.
> 2.) The network node load is never under 3.5 -- 4. This seems to be 
> the case if we are doing 10mbit or 800mbit.
> 3.) Network performance is unpredictable at best.
> Are we doing something wrong ? Have I missed something here ?
> Would moving to "tenant_network_type = vlan" make me happier and if so 
> what do you think is the best way to do that ?
> Any input is appreciated as this is getting to be a pretty major issue 
> :) (happy to engage freelance consultants to help :))
> Thanks,
> _______________________________________________
> Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
> Post to     : openstack at lists.openstack.org
> Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130913/28de1bd0/attachment.html>

More information about the Openstack mailing list