dberg35 is right, you have several links in the chain that aren't addressed in your question. Circuit redundancies, hardware redundancies, application redundancies.
Circuit redundancies- How are the sites connecting now to HQ? private WAN (T1, MPLS, MetroE, etc...), VPN across the Internet (DSL, T1, MPLS, Cable, etc...) Do you have backup connectivity available?
Hardware redundancies- multiple servers in cluster environment like dberg35 mentioned following Microsoft's high availability best practices. Also things like power protection, etc..
Application redundancies- would get this from the clustering environment, but there is also Microsoft's hosted solutions. Moving all your mail to an Exchange hosted provider that has all of the redundancies already built in. Microsoft also has some hosted services which Forefront use to be. MX level virus/spam filtering and mail archiving. Having this purchased would filter the mail, then send it to both the hosted archive and to your internal Exchange server. If your Exchange server became unavailable, then the users could log into the hosted archive and be able to use it to send and receive mail till your local came back online (at least that's how it was explained to me a few years ago when Microsoft first bought Forefront).
Of course, you pay a nice premium for most of these technologies.