SPARC Enterprise systems continue to operate even in cases of component failure. Basic components (HDD, Power Supply Units, Fans, etc) are redundant and hot-swappable. CPUs can be dynamically downgraded.
Basic components such as HDD, Power Supply Units and Fans can be redundant in standard configuration, assuring continuation of system operation.
These functions minimize planned system downtime - needed for system maintenance or upgrade work.
SPARC Enterprise supports hot-swap and hot-plug functions enabling non-stop system operation (24 hours a day, 365 days a year). Faulty components can be replaced and system resources added, without stopping system operation.
Basic components (Fans, Power Supply Units, HDD, etc.) support hot-swap.
CPU/Memory Boards, consisting of CPUs and memory DRAM, can be replaced and expanded using the Partitioning and Dynamic Reconfiguration functions. This function is supported on SPARC Enterprise M8000 models and above.
PCI cards, including fibre channel and LAN cards support hot-plug (*1). This function is supported on SPARC Enterprise M4000 models or above.
*1: For Hot-Plug availability, refer to PCI-card specification.
Faulty memory chips with frequent one bit memory errors downgrade in 64KB units. HDDs, Power Supply Units and Fans can also be downgraded.
Faulty SPARC64 VI/VIIVII+ CPU are isolated and system remains running, assuring system operation continuity.
The faulty memory chip in frequent one bit memory error downgrades in 64KB units. HDDs, Power Supply Units and Fan can be downgraded.
During system start-up, any faulty components are isolated from the system.
Initial diagnosis during system power-on (before OS boot) checks connectivity of CPU/Memory Board and crossbar, and CPU , memory and PCI card installation. CPUs are then activated and memory and PCI-bus diagnostics are performed. A faulty component found at this phase is isolated.
All errors found at the diagnosis phase are notified to the system administrator via the system console.
In SPARC Enterprise M8000 and M9000, dual crossbars connect between CPU/Memory Boards and I/O Boards. Faulty crossbar connections are automatically isolated and the system resumes operation after rebooting. This minimizes system impact from crossbar failure.
In SPARC Enterprise M9000, a failure in one of the dual system clocks triggers a switch to the other clock and the server is rebooted.
Server power can be supplied from dual power sources. In datacenters with a dual power supply system, even if one power supply fails or has an outage the another power system assures power supply to the server.