Coupling Facility Failover

Cluster Services environment failover to a secondary CF in case of hardware fault is not supported. Disconnect by failure & reconnecting to a secondary CF (as defined in Structure policy) is by a fault and recovery using the ADABAS WORK datasets). The failover in this case ,will cause at least some disruption to users who have active UQEs. We would like to propose a less disruptive failover mechanism. This may involve:1) Suspending nucleus activity while a cluster recovery takes place using the WORK datasetsdsn CF or 2) Using ‘User Managed Duplexing’ to maintain a secondary image of the active CACHE and LOCK structure for switchover purposes at fail time. or 3) Using non-fulltime recovery (with again some form of user duplexing) such that at failure time the content of the primary CF structures are copied over to the secondary CF structures while the nuclei are temporarily suspended, allowing normal processing to resume without the need of a WORK dataset online or offline autorestart.

This is R&D Requirement number: 1878