Compare Data Domain

PC Replication / Remote Office Backup Software

For a very small remote office data set, even the entry price of a Bandwidth Optimization box (WAFS, WDS, WAAS, etc.) can sometimes be too much to consider. Here, since the data is small, all that might be needed is a good way to send only changes in files to a hub, based on software within the existing server.

For sites that have ~50GB or less, there is a long list of desktop / laptop / remote-office backup software packages, some of which do deduplication of the data they back up. These have challenges.

• Rip / replace. They typically require rip/replace of existing backup software. Sometimes backup software vendors themselves will offer these products, but as add-ons. They have different mechanisms, planning, and administration tools than their core products.



• Do not scale well. While they work in small environments, these alternatives do not typically scale well to larger datasets or larger sites. For example, deduplicating backup software typically has very poor adaptations for larger files such as databases; it either runs very slow per stream or does negligible deduplication. It may have trouble streaming to tape at speed, because it was not designed for fast read streams. As scale increases, it may require enormous amounts of hardware.



• Deduplication silos. Any other applications, or other clients, will not be able to reduce their redundancy against the data backed up with local deduplication in the backup app. In most other parts of the technology world, when infrastructure can support multiple applications, fundamental functions such as deduplication have ended up in the infrastructure, not in applications.

PC replication software, by using block differencing instead of deduplication, uses about as much bandwidth as an incremental backup in a given day. Backup software that does 'incrementals forever' uses a similar bandwidth envelope (or more). Data Domain systems typically use 80% - 90% less bandwidth than these approaches. Since WAN bandwidth is the biggest cost element in remote office data management, this is a big deal.

Replication is also homogeneous: it requires the same system type on both sides of a link. Backup is heterogeneous; it can support many different client types. If a site is heterogeneous, backup is much more consolidative.