Recovery considerations

Resources should be defined to LPs so that any hardware failure has a minimal impact on the remaining active LPs.

For example, the failure of a physical core can cause the temporary loss of any logical core that was dispatched on the physical cores. In many instances, recovery of a logical core that was running on a failed physical core will take place automatically when an available spare physical core is dynamically brought into the configuration. Also, PR/SM is often able to transparently re-dispatch a shared logical core on a different physical core even when no spares are available. If a logical core is still lost, the LP owning the logical core can continue operating if it was running on an LP with at least two CPs dispatched on different physical cores, and if the control program that is active in the LP can recover from CP failures.