On the hardware side, you could invest in clustered hardware for your most critical servers. On the other hand, you could get your backup/restore plans down cold.
A DR plan and options for H/A really depend on what sort of data/application you are trying to keep available. Some types of data require more expensive solutions than others.
Web servers can utilize Network Load Balancing so that you can take individual servers offline without production downtime.
SQL servers can do log shipping to a standby server.
Mail servers can typically have a backup server handy to either process mail or to restore mailboxes to.
And then there are lots of software solutions that can be thrown into the mix, like Veritas' Bare Metal Restore which allows you to restore a backup to a system that has different hardware.
What sorts of systems are we talking about?
ShackDaddy