Sepaton is claiming an impressive 50:1 compression ratio by combining hardware compression with its de-duplication technology. Backup data is reduced to as little as 4 percent of its original size by de-duplication and then halved in size by the compression hardware. Customers can store more months of backup data on disk and obtain the benefits of fast, disk-speed restores for longer.

Backing up data to tape results in offline tape cartridges. Restoring such data takes time because the tape cartridge has to be located and mounted, and then the data found by reading through the tape reel sequentially. Backing up data onto a virtual tape library (VTL), a product in which disk files mimic tape backup sets, is faster than backing up to tape. Restoring data is also faster than tape because the disks are online and the files can be found by random rather than sequential access.

However, backup files take up a lot of room on disk and de-duplication, the finding and elimination of sub-file level bit strings or block groups across files, can shrink the footprint to ten percent, five percent or even less of its raw size, enabling much more backup data to be held on disk.

Such de-duplication needs a lot of processing power. Data Domain and Diligent de-dupe incoming data, slowing backup speed, while Sepaton and other suppliers de-dupe after the data has been backed up to the VTL so as not to increase backup time.

Now Sepaton has added an Hifn DR 1000, a PCI Express bus hardware compression card, to its S2100-ES2 VTL so as to compress a file of already de-duped data. The LZS compression used, eliminating repeated bytes within a file, typically results in a 50 percent reduction, meaning an effective compression of the raw data to 2 percent of its original size.

Sepaton's CTO, Miklos Sandorfi, said: "Sepaton is the first data protection provider ... to achieve industry-leading data reduction ratios with no performance impact to the backup and recovery process."

EMC/Avamar claims a similar 50:1 de-dupe ratio but by de-duping branch office backup data, collecting it centrally and then de-duping it again. Diligent, FalconStor, Data Domain and other de-dupe suppliers claim ratios of around 20:1. Like car miles per gallon rates though, the actual number can vary with conditions.

Data Domain's MD for EMEA sales, Kevin Platz, made the point that de-duping data at ingestion time means that you can replicate it to a disaster recovery site immediately. The data at the recovery site is then in synch with the primary site's VTL and there is no data loss if a disaster hits the primary VTL.