This blog series is divided in the following parts.
- System Center VMM 2012 R2 Bare Metal Deployment with Converged Fabric and Network Virtualization – Part 1 Intro
- System Center VMM 2012 R2 Bare Metal Deployment with Converged Fabric and Network Virtualization – Part 2 Components
- Host Groups
- Logical networks and IP Pools
- VM Networks
- Uplink Port Profiles
- Virtual Port Profiles
- System Center VMM 2012 R2 Bare Metal Deployment with Converged Fabric and Network Virtualization – Part 3 Logical Switch
- Logical Switch
- Host Profile
- System Center VMM 2012 R2 Bare Metal Deployment with Converged Fabric and Network Virtualization – Part 4 Deployment
- Consistent Device Naming
- Deep Discovery
- Bare Metal Deployment
- System Center VMM 2012 R2 Bare Metal Deployment with Converged Fabric and Network Virtualization – Part 5 Compliancy and Remediation
- Verify settings on Host
- Compliancy and Remediation
In this part two Hyper-V hosts will be deployed. These hosts do not have any Operating System applied. A prerequisite for the Bare Metal Deployment process is that the hosts have their base management controller configured with a user account and an IP address, that is accessible from the System Center VMM 2012 R2 management server. In the deployment process you need to select a RunAs account that will connect to the base management controller, therefore it is recommended to configure the same set of credentials for the base management controller on each server. The host that you want to deploy should have network boot configured before disk boot in its boot order. You can change these settings in the BIOS.
In this blog series two logical switches, consisting of two physical NICs each, are configured on a host. The physical NICs in the servers should be connected to the correct VLANs. This might mean patching physical cables from a rack server to the switch or configuring virtual cabling in a blade chassis. No matter the configuration you need to know what networks are connected to which physical NIC. There are two possibilities to map the correct physical NICs to the logical switches. More recent servers support consistent device naming (CDN) that will consistently name NICs in the BIOS. Windows Server 2012 and Windows Server 2012 R2 can read this information during deployment and name the NICs in the Operating System the same. The physical NIC port numbers in the Operating System will match the numbers on the physical NIC interfaces. For servers that do not support consistent device naming, System Center VMM 2012 R2 provides a feature called Deep Discovery. This step in the bare metal deployment process allows you to get information from the server by leveraging a PXE boot file provided by VMM. Depending on the configuration of the logical switches you will use CDN, Deep Discovery or both.
Consistent Device Naming
If you server supports consistent device naming you can verify the CDN settings in the server BIOS. For this blog I’m using a C7000 chassis with a couple of G8 blades. In the advanced settings of the BIOS there is an entry Consistent Device Naming.
Enabling support for CDN in the BIOS allows the Operating System to read the names from the BIOS during deployment. When the installation of the Operating System is complete the names of the Physical NICs are consistent with the port numbers on your rack server or blade.
Not all servers support CDN. If your host does not support consistent device naming the wizard provides you with an option to run a deep discovery. This optional step will boot the host using the base management controller. We configured the host already to boot from the network (PXE boot) before it will boot from local disk. Deep discovery provides a Boot file leveraging the WDS server.
The boot file will perform an inventory of the physical NICs (Name, Bandwidth, MAC address). This information can then be used to map logical switches to the correct Physical NICs. Deep discovery will take a couple of minutes to complete.
Bare Metal Deployment
To add a new host to System Center VMM 2012 R2 with the bare metal process, open the Fabric tab, select Add Resources and choose Hyper-V Hosts and Clusters. This will start the Add Resource Wizard. Select Physical computers to be provisioned as virtual machine hosts in the Resource location screen. In the Credentials and Protocol screen select the RunAs account that we created for the base management controller. For HP ILO the default protocol Intelligent Platform Management Interface (IPMI) on port 623 is correct.
When you want to assign an IP address from an IP Pool to the Management vNIC on your hosts, System Center VMM 2012 R2 must know the MAC address the underlying physical NIC that is used for PXE. CDN will provide the names of the physical NICs in advance, but not the MAC addresses. So either you enter the MAC address manually or use deep discovery to retrieve the information.
The Discovery Scope screen allows you to enter a single IP address, an IP subnet or an IP range that will be used to connect to one or more base management controllers. The single IP address option allows you to skip the deep discovery process. Verify that the checkmark to skip deep discovery is cleared.
Should you choose to install multiple host (by selecting IP subnet or IP range) you can select or deselect the deep discovery process in the following screen Target Resources.
In the provisioning Options screen you can select to which Host Group you want to assign the hosts (in this example the host group Tenants). You also need to select the Host profile that we created in part 3 of this blog series (Hyper-V Host).
The Deployment Customization screen is where all the pieces of the puzzle fall in place. This is where the actual hardware is mapped to the logical switches.
You will have to wait until the deep discovery process is complete. You can already configure the computer name. When deep discovery is complete you can adjust Network Adapter settings of each host.
Consistent Device Naming configured in the Host Profile (and supported by the host) will map the Physical NICs to the correct logical switches.
Deep discovery provides the MAC addresses of the Physical NICs in order for you to map them to the correct logical switches. If you used deep discovery without CDN map the desired two NICs to the logical switch Management and two other NICs to the logical switch Tenants. To map a physical NIC to a logical switch select the … box next to the Physical NIC that you would like to map.
As noted in the previous part of this series System Center VMM 2012 R2 Preview will not set the port classification correctly. Open the settings of each virtual adapter and verify that the appropriate port classifications is set.
When the mapping of the network adapters is complete you can optionally specify to which disk the operating system will be applied. If you do not specify a disk on the disks tab, the first disk volume that System Center VMM 2012 R2 has access to will be used for the OS deployment.
When you are deploying multiple hosts (which might make sense considering the work to prepare for the bare metal deployment process) you must repeat the steps for each host if you are using deep discovery. So if you are buying new hosts make sure your they come with CDN. It will save you a lot of time and physical NIC mapping, especially if you are deploying multiple hosts.
Completing the wizard will start the bare metal deployment process on the hosts you selected. You can track the process in the jobs tab of the System Center VMM 2012 R2.
Take note that during the creation of the logical switches on the hosts the bare metal deployments process will wait for two minutes. Depending on the performance of your hardware and network a new host will be completely deployed and added to System Center VMM 2012 R2 in about 25 minutes. For those who have tested this process in System Center VMM 2012 SP1 and have failed to complete the process successfully every time I have included the screenshot with the job completed showing the creation of the logical switches in the job log. It really does work!!
In the next and last part of this series we will verify the settings on the deployed hosts and perform management of the logical switches with compliancy and remediation.