High Performance with Snapshots
Snapshots Reinvented!
The XIV system's snapshot offering stands above those of other storage systems thanks to its diverse capabilities and the fact that they are fully applied without any degradation in system performance.
The system's revolutionary high performance with snapshot abilities is made possible by several innovative technologies:
- Innovative meta-data design. Snapshot creation is a practically zero-time command in the XIV system, because the size of the metadata used in the process is truly minimal. In fact, the system requires meta-data only per the total differences between the snapshot and the volume (or between the snapshot and other snapshots). This quality makes it possible to take a snapshot of a volume in practically zero time, regardless of its size.
- Redirect-on-write. Rather than the standard copy-on-write mechanism implemented when a write request modifies data slots shared by source and snapshot, the XIV system implements an innovative approach called "redirect-on-write." The traditional copy-on-write approach involves three disk-seek operations, whereas the XIV system's redirect-on-write involves only two such operations.
- Inter-module copying. The sophisticated algorithm that the XIV system uses to distribute data ensures that redirect-on-write is performed entirely within a module, with no intra-module communication. As a result, a truly huge bandwidth – the bandwidth between CPU and disk drives in a module – is put to the service of the redirect-on-write operation, and no common bottleneck is able to arise. In contrast, traditional storage systems perform copy-on-write between disk shelves through a common controller, connected via FC.
- Concurrent copying in all modules. The redirect-on-write operation is performed on all modules concurrently, load-balanced perfectly by the XIV system's scheme of spreading all volumes over all disks and modules. This provides huge power and full scalability to snapshot performance.
