Why shouldn’t companies be able to retrieve and play back data anytime?
Personal video recorders (PVRs) are changing the way we think about watching TV. TiVo, one of the early pioneers of PVRs, describes its PVR solution as "Automatic recordings of your favorite shows without the hassles of videotape or timers. Watch what you want, when you can!" The idea behind PVRs is to simplify the process of recording so that you can easily retrieve what you want, when and where you want it.
What would happen if we applied these design goals to an enterprise-level backup-and-recovery solution? What features would this "enterprise data recorder" (EDR) need to have, and can you create it with current technology?
The first thing the EDR would need is fast restore capability. The ability to quickly store and retrieve data in random access mode is a must. The price of Networked Attached Storage (NAS) solutions is coming down so rapidly that the cost of disk space is approaching the cost of tape—about 1 cent per megabyte. So, our EDR would use disk as its secondary-level storage.
Second, we need the ability to restore at a granular level. According to the Strategic Resource Group (SRG), more than 98 percent of restore requests are for partial restores of files. For example, if you're using Microsoft Exchange Server, your users typically ask to restore a few deleted messages from a few weeks ago rather than ask you to restore the entire Exchange mail store. Or they might ask you to restore a specific Microsoft Office file rather than an entire user directory. Our EDR needs to provide granular-level recovery—down to the message, file, attribute, or record that you want.
Third, we need to be able to recover data at any point in time. Forget scheduling backup windows at night when the files are closed. That's like saying a PVR can't record "I Love Lucy" while it's playing on TV. We can throw the scheduling problems out the window because our EDR will have the ability to provide snapshots of the data or a replicated copy of the data at any point during the day. This capability is possible with today's snapshot and replication technologies. For example, most NAS devices come with snapshot technology that accommodates differential snapshots of data at certain points in time. With Windows Server 2003 Volume Shadow Copy Service (VSS) technology, applications such as Microsoft SQL Server and Exchange are becoming snapshot-aware, so they can assist in producing clean snapshots of their application data upon request. Replication technology lets you store a duplicate copy of the data across a WAN in real time. Replication and snapshot technology would let our hypothetical EDR record data from 100 remote offices to one central location and let us recover any data to those locations from the centralized EDR.
Fourth, we need an easy way to automatically archive old data. Users keep old files for a long time. Our EDR needs an automated way of retrieving old data from primary storage. For example, if you archived an old email message with an attached 5MB Microsoft PowerPoint file to the EDR, users would then see a small "stub" email in their Inbox with an icon letting them know the EDR archived the message. To get the old email back, they simply click the stub file and the recovery happens automatically. Within a few seconds, the user has the old message back with the attached PowerPoint file intact.
Finally, we need the EDR data to be virus-free, totally secure, and available all the time. Therefore, we need to be able to run standard virus scanners and other storage-management utilities. Our EDR needs to be fault tolerant and replicated to an offsite location in which we can use a third-level storage device such as tape or DVD to store an offsite copy. Because our EDR has clustering and replication technology built in, we can always fail over to a secondary EDR if the hardware or software fails.
Is this dream system feasible? Yes. With a combination of NAS, replication software, backup software, and the snapshot capabilities of VSS, you can create an EDR-like solution today. I've encountered vendors who have built EDR-like systems for their customers. After the early adopters have run these EDR-like solutions in production, storage vendors will package this functionality into a nice bundle and make it available to everyone. Restoring any data in your enterprise from your EDR soon will be as quick and easy as replaying an episode of "Star Trek" from your PVR—without the hassles of tape and timers. Live long and prosper.