One of the cool new features of vSphere 5 is Auto Deploy. This tool enables the stateless booting of ESXi hosts. No longer a installation to disks is required to boot the ESXi hypervisor. Instead the hypervisor is booted via PXE boot into the memory. The ESXi image is booted onto a bare-metal server, so in case the server powered off the “installation” and configuration is also lost. Of course the next time the bare-metal server is booted up again, Auto Deploy makes sure that the stateless image is again deployed via PXE into the memory of the server. So far, so good. If of course the Auto Deploy tool is available to provide the deployment of ESXi images.
And this is the question that always pops up when discussing Auto Deploy in a vSphere design: “What if the Auto Deploy tool isn’t available? Then I can’t boot my ESXi hosts anymore!”
That is true, but fortunately there is a way around that. Daniel Hiltgen, Senior Staff Engineer at VMware, explains how to architect your Auto Deploy environment and make it highly available. This also prevents the “chicken-egg” problem. So have a quick look at this short, but interesting video!