Reliability is in the core of Compuverde - designed to provide Telecom grade availability of 99,999% or better (five nines). This is achieved by the symmetry of the architecture and by keeping the system core small, clean and efficient.
- Virtual IP
- Cache Replication
- Erasure Coding
- Self-Healing Storage Cluster
A failover mechanism makes sure that all the nodes in a storage cluster appear available at all times, continuing to deliver service when a node is taken down for upgrade or in the event of a failure. This is done by acquiring the IP address of the node going offline so that another node can step in. The process is automatic and fast, and the cost is a temporary decrease in performance until the affected node is ready to step back in.
All writes can be mirrored to two nodes in case one node goes down before the content is safely written to storage. Combined with Virtual IP, the next node will continue the write process while the client is automatically redirected to the next node.
Compuverde is not reliant on RAID, as data redundancy is built into the system core. Instead of striping across hard disk drives, we stripe across storage nodes or locations. By using erasure coding, we can reduce the footprint compared to mirroring. For instance, to allow 2 nodes in a storage cluster to simultaneously fail, 3 copies would be necessary – increasing the footprint 200%. With erasure coding 12+2, only 17% added space is necessary – a dramatic 92% reduction. Any two nodes are still allowed to fail, whether they contain the original data or the redundancy, or a combination of both. The process is automatic, 100% accurate and seamless to the client. No "hot spare" or rebuilding of volumes are necessary. Some free space is necessary to allow the storage cluster to self-heal.
With erasure coding 9+3 and 10+6, any three or six storage nodes could break down without losing data or service. 33 or 60% added space are required for this.
Self-Healing Storage Cluster
The self-healing functionality in a Compuverde Storage Cluster will automatically detect defects and start recreate missing data (redundancy) and balance the resources onto available storage. This way, no rebuilding of entire volumes will be necessary and any vulnerability window is kept to a minimum.
Please note: Redundancy is a fail-safe measure, not a substitute for backup.
Read more about our scale-out storage solutions?
vNAS Hyper-Converged Storage