The scouts say something about preparation. I can’t remember what it was but the idea is that it’s better to have done the work to be prepared for a disaster rather than save some time and then get bit*h-slapped by the disaster. For example, what do you do when something nasty happens to your Active Directory Forest? Microsoft posted a guide on this.
I can’t recommend having a lab for your production network enough. Get yourself a TechNet account for this. Set up a single server running something like Hyper-V and create an internal network that matches your production network (or networks). Do a P2V of things like some of your DC’s and other critical systems. This gives you an identical copy of your production system. You may need to do an AD metadata cleanup to remove the DC’s you don’t P2V. You can easily use it for test and development from then on.
I used to do that in the past. I spent 80% of my time on the test systems. For example, when redesigning our AD delegation, I scripted it all using DSACLS. I tested it over and over. When we were ready for production I simply ran the scripts on the production network. Days of work on the test system and minutes of work on the production network with predictable results. We did the same with GPO work and SMS 2003 software deployment.
I’d definitely recommend you do this with processes such as AD disaster recovery testing and for testing backup/recovery.