Dell Unity: How LUN trespassing takes place in Unity array.

The LUN trespass operation occurs for several reasons. From the Unity Side, a LUN trespass occurs on either a Service Processor (SP) failure event for one of the SPs or a host using its non-optimized designed path. When this occurs, the owned LUNs are trespassed to the other SP.

Unity leverages Asymmetric Logic Unit Access (ALUA) for host access. This enables multiple paths to be active for I/O, but some paths are optimized while others are non-optimized. The paths that are designated as optimized are connected to the SP that owns that particular resource. If I/O is sent down the non-optimized path, the peer SP leverages the internal Common Messaging Interface (CMI) bus to redirect the I/O to the SP that owns the resource. This operation is transparent to the host and removes the need to trespass the resource to complete the I/O. This assumes that best practice for connecting a host has been followed - two HBA's per host, two switched and two SP ports per SP. This is configured for four paths - see second image below.
 
Example: Here are different scenarios for LUNs being trespassed to the other SP "initially unowned one"
 
Scenario1:
In the image below, if SPA goes down, its owned LUNs are trespassed from SPA to SPB. Once SPA returns to service, the ownership of the LUNs are redirected back to the initial owner -  Trespassed back to SPA. 

  

Scenario2:
If many I/Os are received down the non-optimized path, Unity will trespass the LUN to the peer SP in order to optimize the data path.
The later mentioned scenario happens when by error the Host sends its data on a wrong path so accordingly Unity will automatically trespass the LUNs to optimize the data path. 
 
This is how Unity avoids the high utilization of an SP that does not own a certain LUN.

Scenario3:
In the image below, if an SP fails, the LUN fails over to the surviving SP and continues to service I/O since it is connected to the same switches. If there was switch failure, the remaining switch provides access to both SPs, eliminating the need to use the non-optimized path. If there was a host failure, the cluster initiates a failover to the other host and brings the application online. Any path failure due to a bad cable or port does not cause any issues since the second optimized path can be used.
 
This configuration can also survive multiple failures, as long as they are not within the same component. For example, failure of Host B, Switch A, and SPA can be tolerated since the surviving components can be used to access the LUN. In this case, Host A can connect through Switch B, and access the LUN that s trespassed to SPB.