The server deployment needs to be straightforward. No-one likes disaster recovery plans that include "Dance naked around a maypole embedded in sand of the purest white while offering prayers to the great gods".
So realistically the question is, what needs to be recreated during a "DR" install?
In an ideal world, nothing. Everything should be configured in such a way that just restoring the files will recreate all configuration required - the obvious and only exception is services.
Now realistically, if someone is backing up "System State" on their servers, and they darn well should be, then when they go ahead and restore, the service configuration will be retained. This gives us complete recovery with the following steps:
- Build a new Windows Server with the same name as the old server. Don't join the domain.
- Restore the System State and all files.
Ahem. So we've got a restored server with all the configuration we had before the fatal crash.
Replacing the Server
Slightly different - we need to run setup and reinstall all the configuration we had before. Ideally the plan here should be to establish a data file during initial setup that we can follow through for the reinstall. If SQL was local, then reinstall it. Prompt for the database backup so we can get the data across. Install services. We will go ahead and fix the group policy we created that deploys the client application too - remove the old source and add the new source for the app. Auto repair here we come!
Reinstalling the Client
This is easy. Uninstall the application from Add/Remove Programs on the client and re-apply Group Policy to that machine (XP hint:
gpupdate /force) then reboot. Voila!