Combine features of the above three models, data sharing technology directly interfaces with the individual processors to enable each processor to cache shared data in its own local memory with full read/write capability (as in shared-memory model) Data sharing maintains data integrity by ensuring cache coherency across all processors and provide efficient cross-system locking mechanisms (as in shared-nothing model)All disk devices are accessible to each processor so that a workload may be scheduled to run on any available processor (as in shared-disks model), instead of being run only on the processors with data affinity (as in shared-nothing model). Example of shared-data system: DB2 on IBM System/390 Parallel Sysplex. |