Deploying RHEV 3.6 pt5 (Networks)

Hi folks,

We’re officially on a roll here and making headway. This is the last major deployment and/or configuration step within RHEV before we start pushing things around. Yes, there are lots of bells and whistles, but in terms of “compute”, “storage”, and “network” resources, this rounds it out. Let’s get to it.

Like everything else so far, deploying networks in RHEV is really straightforward. It can be done from the “Networks” tab or the “Data Centers” tab, but know that ultimately you’ll want to do some additional configuration within the “Logical Networks” area of the “Clusters” tab. I’ll explain in a bit.

When you initially create a logical (virtual network), it is assigned a name, a description, and optionally a comment. VLAN tagging is fully supported, as are Jumbo Frames of any size. And if you don’t want VMs to traverse the particular logical network that you are setting up, then uncheck the “”VM network box” – you’ll see it in the video below. This is very useful if you just want to use the network for management or storage traffic.

In the “Logical Networks” area of the “Clusters” tab, there are some additional parameters of note. Because you can have multiple logical networks per cluster, you may need and want to determine which networks can carry which traffic. If you want to limit your management traffic to the onboard 1GbE interface, it’s possible. If you want to guarantee that display and migration traffic are on logical networks that will be assigned to VLANs carried by a 10GbE bonded interface, that can be done.

The other configuration option of note is the “Required” check box that you’ll see in the video. It simply means that the logical network is required in order for VMs to operate. For example, in a 4 node RHEV cluster, if one of 2 interfaces on node 4 drops and it happens to have a “required” logical network, that node would be removed from the cluster. The VMs would be evacuated from that node and migrated to other nodes in the cluster and VMs would be prevented from migrating to the offending node. Ultimately you don’t want VMs operating at one level, then migrating to a node with a lesser capability or service.

Once the logical networks are configured, the last step is to assign them to interfaces on the hosts (hypervisors). Understand that under the covers, the interfaces are Linux bridges. Because of this, you can assign multiple logical networks to the same interface. I will highly recommend that if you do this you make use of VLAN tagging so that you can properly segregate your traffic.

Once all of the hosts are configured with the new logical network(s), VMs can be created and assigned to that logical network.

Enough talk, let’s catch the video (best watched in full screen):

Again, this is all very straightforward and meant to set things up for bigger and better things down the road. Comments and questions are always welcome.

Hope this helps

Captain KVM


Agree? Disagree? Something to add to the conversation?