In PRIMECLUSTER, the following methods can be selected for taking over data between cluster nodes:
A shared disk method
With this method, data is stored in a shared disk unit.
You can use the following shared disk units:
A shared disk unit connecting to multiple servers through SAN (Storage Area Network)
This is suitable when data accessibility performance or data availability is especially important, or data capacity or data scalability is required.
NAS (Network Attached Storage) unit
You can build a cluster system at a relatively low cost.
Mirroring among servers methods (Linux)
With this method, data is stored in local disks (internal disks) on each node and those internal disks are mirrored through a network with the mirroring among servers of Global Disk Services (hereinafter GDS).
You can build a cluster system at a low cost because an expensive external storage is not required.
This is suitable for a small-scale system that keeps and updates less data.
Figure 1.3 Taking over data between nodes
