Hardware partitioning divides server resources into multiple server entities where OS and applications work independently(*1). SPARC Enterprise partitioning functions are highly reliable and flexible(*2).
SPARC Enterprise can be divided at an individual CPU level into physically multiple independent server entities. Any workload increases or a failure in one partition has not effect on the other partitions.
Hardware partitioning is available with SPARC Enterprise M4000 and above.
*1: Also referred to as a "domain".
*2: Hardware partitioning and Extended partitioning supported on PRIMEPOWER are collectively referred to as partitioning.
SPARC Enterprise uses two basic partitioning methods, one based on CPU Memory Board units and the other on CPU chip. These partition types can be mixed in the same server and hardware errors in such partitions have no effect on other partitions due to the ability to isolate partitions with a fault.
There are no performance overheads due to hardware based partition control.
Reliability and efficient resource utilization are well balanced, due to the fine-grain hardware partitioning down to single CPU chip.
Hardware partitioning provides superior business continuity over software and logical partitioning. Single points of failure have much less effect than with software and logical partitioning because errors in a hardware partition never effect other partitions. OS failure in software partitioning, and firmware failure with logical partitioning, can halt the entire server.
SPARC Enterprise organizes the use of multiple partitions. Sudden server workload increases are resolved by adding pooled resources to the relevant partition to match the workload change.
SPARC Enterprise XSCF enables partition management (including configuration) and Dynamic Reconfiguration operations. User accounts separated per partition provides high security and manageability.
Partition operations such as configuration and hardware resource transfer can be made by signing on to the relevant partition as well as by XSCF. Automatic resource transfer between partitions can be made with the sign-on method similar to PRIMEPOWER.
DR operation by XSCF is suitable for data center operations and access between partitions are strictly prohibited due to high security requirements.