Networking Options with LTM VE

If you haven’t yet downloaded the  BIG-IP LTM VE trial, I highly suggest you do.  It is a fully-functional LTM, rate-limited to 1Mbps throughput.  If you’re not familiar with virtualized environments, hopefully this blog will fill in some blanks for how to get started on the network front.

Getting Started

Before downloading your VE image, you need to choose what virtualization environment you’re installing into.  The supported options in the type 1 hypervisor are VMWare ESX version 4 and ESXi version 4.  For the type 2 hypervisor (requiring a host OS such as linux or Microsoft Windows) the supported option is VMWare Workstation 7, which offers a 30-day free trial that I recommend you give a shot, or for those with experience on VMWare player, that also will suffice if you are at version 3.  Note, however, that VMWare player is not supported by F5. 

Hypervisor Type 2 Options – VMWare Workstation & Player

In LTM VE, you have three interfaces—one managment and two data (1.1 & 1.2).  On the Workstation/Player products, you specify in the virtual appliance settings how the interfaces will connect.  You can specify any of the following interface types:

• Bridged – Allows access through your physical NIC to participate in the local area network.
• NAT – Allows access through your physical NIC, but utilizes your machines IP and translates for VM traffic.
• Host Only – Networks are defined locally in virtual nics that have no significance outside your locally defined virtualization environment.

With the Workstation product, there is a Virtual Network Editor application where you can define the networks your virtual appliances will use, as well as setting dhcp options, etc.  The player doesn’t have this application, and doesn’t give the custom option in the GUI interface, but the settings can be configured manually in the appliance configuration files (shown below).  To get started quickly, I bridge the management interface so I can download directly from the management shell.  I use a host-only interface assigned at layer3 on both my laptop and the VE image so I can run test traffic against my iRules for syntax and functional checking.  I have a virtual appliance on a layer2 network (layer3 for VE and the server appliance, but there isn’t a layer3 interface for vmware itself) between it and VE so I can pass traffic from my laptop through VE to the vm server and back as necessary for testing.  A diagram detailing this is shown below to the left of the matching configuration options set in the virtual appliance files.

# MGMT NETWORK
ethernet0.present = "true"
ethernet0.virtualDev = "vlance"
ethernet0.addressType = "generated"
ethernet0.connectionType = "bridged"
ethernet0.startConnected = "true"
# INT 1.1
ethernet1.present = "true"
ethernet1.virtualDev = "e1000"
ethernet1.addressType = "generated"
ethernet1.connectionType = "custom"
ethernet1.startConnected = "true"
ethernet1.vnet = "VMnet1"
# INT 1.2
ethernet2.present = "true"
ethernet2.virtualDev = "e1000"
ethernet2.addressType = "generated"
ethernet2.connectionType = "custom"
ethernet2.startConnected = "true"
ethernet2.vnet = "VMnet2"

I think in order to take advantage of route domains on the workstation product, you’d need a couple virtual appliances in different vmnets that are only layer2 aware.  Still, there are plenty of possibilities with apache vserver configurations if you have the memory to spin up a virtual appliance in addition to the BIG-IP LTM VE.

Hypervisor Type 1 Options – VMWare ESX 4/ ESXi 4

In ESX/ESXi, it’s both more complicated and more simple.  Yeah, I said that.  The assigning of interfaces is trivial, as there really isn’t a concept at the virtual appliance level of bridging, natting, or host only.  The ESXi platform has an underlying virtual switching infrastructure where all the science of networking is configured.  You can teem your nics and run all your vlans across them, or you can segment by function.  When deploying the .ova image to ESXi, the only interesting questions are what datastore you will use to house your VE image and what networks to apply to the VE interfaces.  Given that you cannot create them on the fly, you’ll need to do some prep work to make sure your interfaces are already defined before deploying the image.

Questions?  We turned on an LTM VE specific forum today should you have any questions regarding installation, network configuration, VE options, etc.  We hope you get as much use and enjoyment out of this release as we do.