ALUA Working Principles
When ALUA works, the host multipathing software classifies the physical paths to disks as Active Optimized (AO) and Active Non-optimized (AN) paths. The host preferentially delivers services to the storage system via the AO paths.
- An AO path is the optimal I/O access path between the host and the owning controller of a LUN.
- An AN path is the suboptimal I/O access path between the host and a non-owning controller.
ALUA Working Principles and Failover in Non-HyperMetro Scenarios
If an AO path fails, the host will deliver I/Os to another AO path.
If all AO paths on the owning controller fail, the host will deliver I/Os to the AN paths on the non-owning controller, as shown in "Path failure" in Figure 6-1.
If a LUN's owning controller fails, the system will activate the other controller as the new owning controller, as shown in "SP failure" in Figure 6-1.
ALUA Working Principles and Failover in HyperMetro Scenarios
When HyperMetro works in load balancing mode, the host multipathing software defines the paths to the owning controllers on both HyperMetro storage arrays as AO paths, and those to the other controllers as AN paths. The host accesses the storage arrays via the AO paths. If an AO path fails, the host will deliver I/Os to another AO path. If a LUN's owning controller fails, the system will activate the other controller to maintain load balancing.
When HyperMetro works in local preferred mode, the host multipathing software defines the paths to the owning controller on the local storage array as AO paths. This ensures that the host delivers I/Os only to the owning controller on the local storage array, reducing link consumption. If all AO paths fail, the host will deliver I/Os to the AN paths on the non-owning controller. If the owning controller of the local storage array fails, the system will activate the other controller to maintain the local preferred mode.