IBM ProtecTIER Deduplication Enhanced For Disaster Recovery

IBM enhanced the disaster recovery capabilities of their ProtectTIER deduplication, adding deduplicated replication from one ProtecTIER to another. The software feature, available now on ProtecTier appliances and gateway product lines, is list priced at $2,000 per TB of data. Until IBM added remote replication, the company relied on the storage array to replicate data or used third party products. With this enhancement, deduplication is tightly integrated with the ProtecTIER product.

Victor Nemecheck, ProtecTIER Offering Manager with IBM, says: "Organizations use network replication to protect their most critical data because the bandwidth required to replicate storage is expensive. With ProtecTIER's deduplication, the bandwidth needed for replication can be reduced by 20-25%, on average, which represents a significant savings."

After data has been sent to the ProtecTIER product and has been deduplicated, it can be written to the local media and sent to a remote ProtecTIER product. If the data is new, then it is sent in its original form. If the data is a duplicate, then only the meta data needed to recover the original data is sent across the network. That is where the bandwidth savings comes in. On the receiving end, ProtecTIER simply writes to its local storage. Administrators can select what data is replicated base on a number of criteria such as backup-set, VTL level, or a whole library. For example, business critical backup data, such as sales information and inventory, can be replicated to a remote location while less business critical data, such as logs, can be stored locally.

Backing up data for disaster recovery often means writing data to tape or some other physical media, labeling the media and them moving it off-site. If your DR site is located far from your home office, that means shipping the media, which can take days. In the event of a disaster, the backed-up data has to be restored. Using physical media is a time consuming and expensive process that delays disaster recovery time. Backing up storage over the network is easier and less error prone. The ProtecTIER product at the remote site is a fully functional appliance or gateway and takes up deduplication tasks when a disaster recovery operation is started and the remote site is turned enabled.

David Hill, Principle with the Mesabi Group, notes: "While what IBM is doing is not unique, it is important. The traditional use of storage deduplication is to efficiently store backup data on disk. The other use of backup data is for disaster recovery (DR). Electronic vaulting, which is sending backup data over a network, is faster and more reliable than physically moving it, but without something like data deduplication to reduce the amount of data sent over the network, bandwidth costs can be too high. ProtecTIER is now a more comprehensive product that supports both everyday operational recovery and sets the stage for disaster recovery."Deduplication does offer a number of advantages, such as reduced bandwidth requirements and more efficient use of the available bandwidth. WAN optimization products such as Silver-Peak's NX appliances can reduce the WAN bandwidth required to replicate data to a remote site. The company works closely with storage vendors such as EMC to ensure seamless integration with storage products, but the data written to disk or tape is not deduplicated. However, it's also true that disks or tape are relatively cheap and adding physical capacity may be more cost effective and more network efficient than paying per TB licensing fee for ProtecTIER's depuplication.

The decision to follow WAN optimization or deduplicated replication is not a clear one. You have to factor in the cost and benefit of WAN optimization alone versus ProtecTIER's deduplicated replication. WAN optimization reduces WAN bandwidth requirements--providing the data is compressible--but then add in the cost for media to store and manage the original data. Compare that to the the increasing licensing cost--licensing is based on capacity--for ProtecTIER 's deduplication minus the savings of not having to purchase more physical media. Your mileage may vary.