Network storage systems having clustered raids for improved redundancy and load balancing

A computing device comprising:

• a first module adapted to determine:
• a first path through a network, the first path coupling the computing device to a first storage controller providing access to a first plurality of storage devices;
• a second path through the network, the second path coupling the computing device to a second storage controller providing access to a second plurality of storage devices different from the first plurality of storage devices, the second storage controller forming a first high availability pair with the first storage controller, the second path including an inter-controller communication channel coupling the first storage controller and the second storage controller;
• and
• a third path through the network, the third path coupling the computing device to a third storage controller providing access to a third plurality of storage devices different from the first and second pluralities of storage devices, the third storage controller being part of a second high availability pair different from the first high availability pair, the third path including a first volume mirroring link coupling the first storage controller to the third storage controller;
• and
• a second module adapted to:
• receive a non-backup write request from a host server, the non-backup write request including data for storage on the first plurality of storage devices;
• determine which of the first, second, and third paths are available;
• select one of the first, second, and third paths based on the determined availability;
• and
• send the non-backup write request to the first storage controller using the selected one of the paths.