Shared storage works by separating the servers that process data from the storage, using various technologies that are beyond the scope of this book (the most common being Fibre Channel and iSCSI). The idea is that two servers with the same storage space can easily act as in active/passive mode (with one node active at any one time), and can quickly fail over the active node.
The simplified structure of shared-storage cluster is as follows:
Note
Note that between the SAN and nodes, there are generally multiple paths for the storage traffic to take, for both fibre channel and iSCSI based solutions, these often involve multiple controllers on the storage unit. This book does not focus on this layer, but multipathing is an essential part of a production setup.
Depending on the solution developed, the service may run on one server at a time (either Node A, Node B, or Node C as shown in the previous diagram) or run on all three nodes with a filesystem...