Q. When should I use AutoDeploy and when I should not?
A. AutoDeploy is designed to give administrators the ability to deploy quickly to numerous amount of machines at a single time using either multicast or unicast deployment if you want to do one at a time.
So for example you have purchased 50 new machines and you want to deploy your standard image. So you setup a lab in an isolated network so you can deploy this new image.
In this case Autodeploy works wonderfully in this new lab environment and it will act as it should.
However setting Autodeploy on in your production environment will switch on the DHCP server on the repository so any machine that PXE boot will get an IP address.
Now what is supposed to happen is that the DHCP server on the repository will not respond in time and so the target endpoint should an IP address from your main DHCP server if its not PXE booting. This works generally very well with Windows based machines.
However what we have found is that some devices that have a non standard IP stack will ignore your main DHCP server and decide that the PXE address range is just dandy and so we have devices appearing on the PXE subnet range when they shouldnt.
Now because they are getting addresses inside the PXE range they stop working on your production network, and its strange devices like switches, routers and even some game consoles. We have had some success with firmware updates with such devices.
However from our manual it states:
"Auto Deploy is primarily designed to facilitate image deployment to new or reformatted machines, booted up in a controlled, isolated range of IP addresses. "
So unless its critical to have AutoDeploy on in your production network then switch it off.