A common concern that students have are the changes made between Exchange Server 2003 and Exchange Server 2007 when it comes to providing service redundancy and disaster recovery.

So, here is my response, at a high level.

First - You need to do what you should do with every single server in your environment, you need to harden the operating system and the application. The easiest way to do that is to use the Security Configuration Wizard (SCW). In order to use SCW, you need to first configure SCW so that it knows about Exchange Server 2007. Luckily, I have already written how to do that here for you.

Second, you need to look at each of the roles and decide whether redundancy is needed for them or not.

The mailbox role is often the most critical because it contains alll of the data for your Exchange environment. You need to decide what is the best method to use for providing high availability for your mailbox servers. I updated this blog just a few minutes ago to include some discussion on Single Copy Clusters (SCC) vs. Clustered Continuous Replication (CCR) here.

The Hub Transport and Client Access Server services are also considered extremely important for most organizations. In order to keep the number of servers down, combining these two roles on the same servers is a pretty common practice. However, the way that redundancy is usually implemented is not the same. Anyway, to make a long paragraph short, just read this blog entry regarding how to configure Network Load Balancing for the HT and CAS roles and it should help you out a great deal.

In order to provide high availability for the Unified Messaging role, just deploy multiple UM servers with the same rules on each.

As far as the Edge server role goes, you can configure one Edge server and then export its rules and import them on a second Edge server to provide support. You should also configure subscriptions on multiple Hub Transport servers. To provide redundancy for messages coming in from the Internet, simply create multiple MX records and point at your individual Edge servers. Provide the same priority level for both and they will nicely round robin.

Third, you should look at Standby Continuous Replication (SCR) to provide a copy of your database at a remote location for disaster recovery purposes.

Last, please, please, please configure a proper backup and recovery strategy and test it over and over again so you are confident that you can recover any server you need to in a little time as possible with as few mistakes as possible.

As always, I hope nothing ever goes wrong with your environment, but just in case it does, keep your resume up to date <G> and be prepared for the worst to happen.