Many companies have a Disaster Recovery plan that is focused on getting their web, application and database servers up and running as quickly as possible after an outage. But what take a lower priority are the email servers. And lesser priority than the servers are the email archives.

When creating a DR plan for data – in this case email data – it is important to specify a Recovery Time Objective (RTO). Ideally, it would be nice to be able to recover your email data from any point in time but that is not always possible.

So you must establish which emails data are most important. All emails are not equally important. Some emails contain non-critical communications both internal and external. But other emails may communicate financial information that is not only critical to the operations of your company but also highly proprietary. Still other emails may be of a legal nature that cannot be lost and must be recovered quickly due to possible impending litigation. Any emails lost or corrupted can have a negative impact on company operations.

So the question is how often should you backup your email archives?

Are scheduled backups every hour sufficient for your company? Or are the emails that flow through your company so vital in nature that your company would suffer if restored emails are even minutes old? In this case you need constant backups.

If continuous backups are required then what backup medium will be chosen? Will it be tape or disk?

It is estimated that 40% of backups will fail. Backups rely on many infrastructure layers such as the network layer, the storage server(s) layer, the tape or disk medium being used and sometimes just plain old human interaction. All are prone to failure.

Administrators have to ensure that all applications and relevant databases have been stopped. And then the last complete backup must be located and loaded. Any incremental backups must be applied and finally the database logs must also be applied. The sum of these actions will take hours if your email archives and databases are the size that most enterprises must support.

I have seen data center recovery operations take hours and hours before end users were able to log back into the applications. So having a Recovery Time Objective (RTO) specified in your Service Level Agreement (SLA) is just as important as having your RTO defined in the first place.