Architecting a vSphere Compute Platform : Planning Host Deployment : 6.1 Preparing for Host Deployment
   
6.1 Preparing for Host Deployment
The ESXi installer, unlike most other operating systems, copies the system image to the installation destination, and therefore does not actually install in the traditional sense. This makes the deployment process very fast, requiring little interaction from the deployment engineer. The source of the system image is typically provided to the server hardware through a CD/USB for local interactive deployment, through an ISO image for remote KVM deployment, or through a PXE boot from one of the server’s network cards.
As of vSphere 5.1, in addition to stateless deployments offered by VMware vSphere Auto Deploy™, a stateful installation of the system image can also be initiated from a vSphere Auto Deploy server through PXE, which is covered in more in depth in the upcoming section on vSphere Auto Deploy.
An interactive installation, the default and most straightforward method for smaller deployments, is a simple routine where the deployment engineer responds to different options and provides the answers, either locally, or remotely through a remote console system, such as HP Integrated Lights-Out (iLO), Dell Remote Access Card (DRAC) or Cisco Integrated Management Controller (CIMC).
However, where the number of servers being deployed best suits a “light touch” approach, a scripted installation might offer several advantages. Employing the use of a kickstart script for the mass deployment of hosts or ongoing deployment of hosts, not only provides a means to fully automate the operating system deployment, but also:
Provides a repeatable and consistent process for ESXi deployment
Provides a standardized build, reducing operational overhead
Speeds up the deployment process
Provides additional post-deployment configuration options
The ESXi answer file (ks.cfg) mimics the syntax found in Red Had Linux kickstart script, and can provide a scalable and consistent approach to host deployment by being practically “hands-free” after the initial effort to set up, test, and customize is complete. Therefore, this type of build environment is ideally suited to large service providers who are potentially deploying hundreds of servers each week.
 
The next consideration required for host deployment is the destination of the system image itself. The ESXi hypervisor binaries can be deployed to several different locations. Traditionally, VMware ESX® or ESXi was deployed to local hard drives often configured as a RAID 1 mirrored pair for improved availability. However, in recent years and for various reasons, such as cost, power, and simplification, there has been a wider adoption of the use of internally mounted USB flash and SD cards. This trend is set to continue with the constant growth in the use of vSAN and VMware vSphere Virtual Volumes™. However, these locations are considered removable by the ESXi installer and are therefore subject to the caveats and considerations for log files discussed previously.
Another destination that is considered remote by the ESXi installer is a boot from SAN (BfSAN) LUN. A bootable SAN LUN can be configured over FC, FCoE, or iSCSI, and the hypervisor can boot from it directly through an appropriate initiator.
The following section addresses the advantages and drawbacks of each system image destination option to provide the service provider the information to make the most educated and up-to-date design decision.