Rethinking backup in the era of the cloud
In the scale of miserable IT jobs, backup ranks up there with building a full rack of razor sharp 1Us. Backup is cumbersome, under-appreciated and you always end up bloody. (See my recent blog post on how RAIN architectures fall short for backups.) In addition to being complex and costly, backup puts a tremendous load on your servers and network as the data gets syphoned through the backup server.
Backup serves two critical purposes: making a copy of data outside the storage system you wish to protect and versioning everything so you can roll back unwanted changes. Many people come to us every week asking for backup in the cloud. Backup equals pain and the cloud equals IT panacea. Therefore, the cloud must be good for backup.
Wanting to use the cloud for backup is typical example of how truly transformative technologies emerge into being. Before the internal combustion engine ushered in the era of the automobile, the horse and buggy ruled the roads.
Our first reaction to new technology is to mimic the elements of the old technology. Powered by an engine, this mechanical horse is the equivalent of backup to the cloud. It looks advanced but has the reassuring familiarity of the old ways. It is also wholly impractical.
Can the cloud make backup obsolete? Curtis Preston wrote a courageous piece on using snapshots and replication as an alternative to backup. NetApp has been trying to move the world from using backups to snapshots and replication: versions and copies. NetApp is a best-of-class NAS but its technology fundamentally depends on storage arrays. For normal storage, snapshots mean local copies. Those copies can then be copied to a second site via replication. In traditional snapshots there is a significant storage overhead to each snapshot that ends up placing a practical limit as to the number of snapshots that can be kept. That means that while in theory traditional snapshot and replication equals no backup, in practice, no one can afford the risk of depending only on a second site and for a limited number of snapshots. Backup to tape or disk has many drawbacks but organizations have continued to use this procedure as the way to save copies of snapshots to multiple locations. A practical snapshot-plus-replication-only solution would need much more than just a second copy and limited snapshots.
At Nasuni we saw an opportunity to completely rethink NAS and, in particular, snapshots so as to completely eliminate the need for additional backups. We built data protection into our NAS. We architected snapshots to be an intrinsic property of the file system instead of an add-on function. A Nasuni snapshot is actually the way that data is stored in the cloud. The images are highly compressed and de-duplicated making the overhead insignificant. Because the file system is based around snapshots, there is no failure case that stops snapshots from occurring. Our snapshots provide a full Disaster Recovery image, as frequently as every hour, forever, and then leverage the cloud to replicate those images to multiple data centers. This is a far cry from traditional NetApp or Linux LVM snapshots. The combination of Nasuni’s unlimited snapshots plus being able to store the data in multiple state-of-the-art data centers offers the first practical alternative to backup ever.
If you want to release some of the pressure on your backup environment, the Nasuni Filer is a NAS that needs no backup. That means that you can host all those Network Shares without ever needing to back them up.
It means backup is built-in.
It means complete data protection without backup.
It means transportation without the horse.
It means a whole new piece of technology for getting your job done.
Come ride our Model T!