Backup-to-cloud presents an alternative to backup-to-tape for moving backup sets off site. WD recommends that all administrators maintain backup sets off site to protect against natural disasters and theft. See How-to-Choose for a discussion of when tape or cloud offers the better solution for off-site storage. Today, 9% of WD Arkeia customers protect backup sets on public or private clouds.
Private and public clouds can be used either as a direct target for backups or as a target for the replication of backup sets. Backups direct-to-cloud have gained some popularity among consumers as a mechanism to protect individual computers. For enterprises and other organizations, cloud storage is generally better suited as an off-site destination for backup set replication.
For some enterprises, the best strategy is to perform backups to disk on local area networks (LANs), and then replicate backup sets to the cloud. This strategy named hybrid-cloud backup shortens the backup window and accelerates most data recovery operations.
A local backup server shortens the backup window because LAN bandwidth is typically 10x or 100x WAN bandwidth. Because a local copy of backups (usually a few weeks of restore points) is generally retained on the LAN, the vast majority of restores can be performed directly from the backup server on the LAN. This is true both for the recovery of individual files and full disk images.
Generally, administrators require recourse to the cloud-stored replicas only if the LAN-resident backup server is unavailable (e.g. due to theft or natural disaster) or more than a few weeks old.
As for backups direct-to-cloud, transfer of the initial backup set to the cloud, and recovery of full disk images, can require several days or weeks over the WAN. For this reason, both the initial backup and full disk recovery of large backup sets may be best transferred on physical media (e.g. ruggedized USB disk or NAS). Subsequent incremental backups or snapshots are transferred over the WAN.
Backup direct-to-cloud can leverage strategies like full and incremental backups, as well as deduplication or file differencing to shrink the volume of backups. Nonetheless, full backups to and disk image restores from the cloud are often impractical because of the long time need to push large data volumes over the WAN.
Data deduplication plays a major role in backup to the cloud because it can dramatically reduce the bandwidth necessary to replicate, and the disk volume necessary to store, backup sets.
Data deduplication is most valuable for backups of multiple computers on a LAN because deduplication identifies and eliminates redundant data across computers. Nonetheless, backups of individual computers to the cloud generally achieve compression rates that are much inferior to compression rates achieved when multiple computers are backed up and jointly replicated to the cloud.
File differencing compression techniques (sometimes called byte-grain deduplication or delta differencing) are not able to identify either redundant data across computers OR similar files with different names. For this reason, file differencing offers a poor solution for cloud storage of LAN backups.
Learn about WD Arkeia network backup version 10 hybrid-cloud backup state-of-the-art features.