During his presentation at the Customer Forum last week Curtis Preston commented that the choice of backup storage technology should not compromise the restore time. His comment was specifically aimed at deduplicating technologies and was a timely reminder that you should always test the time to restore when evaluating deduplicating storage. (He also made the valid point that not all deduplication is the same and you shoud always evaluate at least two dedupe devices and compare the space savings.)
His comments reminded me that the whole reason we do backups is to be able to restore and that we should never lose sight of that all important RTO target. Technologies such as client side dedupulication and synthetic full backup are very good ways of achieving a backup when bandwidth is limited but you need to remember that recovering a server or application over a narrow pipe is going to take a long time.
When it comes to mission critical servers with large amounts of data and short RTOs there really is no substitute for SAN Clients, fast LAN connections and SAN Media Servers that allow you to do true full backups and restore them quickly in the event of a disaster - you can always use server side deduplication to reduce the amount of storage space used.
Use client side dedupe and synthetics for non-critical server backups by all means but there's no substitute for a fat pipe when the RTO is tight.
