The optimal backup solution: it's now within your reach

Computer Technology Review, Oct, 2004 by Jim McKinstry

There are a variety of ways to perform snapshots:

Split-mirror snapshots create an instantaneous second copy of data that can be used at a later time to recover data: this is accomplished by mirroring disk drives and then breaking the mirror. It is traditionally done within the disk subsystem or from the host using some sort of volume management software, such as VERITAS Volume Manager. Today, SAN appliances like FalconStor's IPStor also perform this function. The major issue with a split-mirror snapshot is that there must be enough storage to accommodate not only the original copy but the split mirrors as well. Assuming that a snapshot is taken every hour and held for 24 hours, a database with one terabyte of disk space allocated to it would need 25 terabytes of usable disk space (production copy plus 24 snapshots). By retaining multiple split-mirror snapshots, the amount of time to recover will be much less than going to a backup system to recover the data.

Snap-copy snapshots, like split-mirror snapshots, provide a complete copy of the data. The major difference is that when a split mirror is initiated, it creates an instantaneous copy of the data, so the data is available immediately. When a snap-copy is initiated, the data is copied to another area of storage, which may take from a few minutes to hours. As with a split mirror, each snap-copy snapshot requires enough storage to hold an exact copy of the original data and can be created on the host (Volume Manager), within a SAN appliance (IPStor) or on the disk subsystem using, for example, Engenio's SANtricity Storage Manager software.

Pointer-based snapshots are not exact copies of the data but a set of pointers that point to the original data. As the original data is written to, the changed blocks are written to the snapshot reserve area and the pointer moved to that block; this process is called "copy on write." Subsequent writes to the original data are not copied to the snapshot reserve area because the original data has already been moved. One of the most attractive aspects of a pointer-based snapshot is that the snapshot reserve area needs just a fraction of the original disk space, since only the changed blocks are copied. Because pointer-based snapshots require such a small amount of space, they can be taken more frequently at a low cost. Pointer-based snapshots are the most robust of the snapshot technologies and can be created on the host (Volume Manager), within a SAN appliance (IPStor) or on the disk subsystem (SANtricity).

Optimal Solution

The optimal solution, which solves a majority of most companies' backup issues, is to implement a SAN with Fibre Channel disk with snapshot technology for production, ATA for disk-to-disk backup and a tape library at a remote location attached via a stretched SAN. A backup would consist of a snapshot of the production data, which would be mounted to a backup server. The backup server would then backup the snapshot to the ATA disk. That backup would then be archived to tape. The copy of the backup that resides on the ATA disk would "age" using the same retention policies that were in place for the local tape copy. The backup that was archived to tape can remain at the remote site and fulfill most companies' requirements for offsite tape storage.