High Availability Requirements

This section summarizes the requirements for high availability.

Through ACPI and OnNow power management capabilities, Windows NT 5.0 allows more control of dynamic configuration changes and power state changes. These features help implementers in handling event-based issues such as lights that are keyed to system failures, pending failures, or system power states.

System includes alert indicators for occurrence of failure

Alert indicators should be provided that indicate hard failure. In addition to visual alerting mechanisms, a design can also provide software alerts such as paging, fax, or e-mail notifications.

The following are required sources of alert indicators for hard failures for systems running Windows NT Server/Enterprise Edition. These indicators are recommended for servers running other versions of Windows NT Server.

• Cooling fan malfunction
  
  
• Disk drive error
  
  
• N+1 power module failure

These sources of alert indicators for hard failures are recommended for all servers:

• Chassis cover open (intrusion)
  
  
• NMI, processor internal error, and time-out of watchdog timer

Hot-swappable drive includes a local disk drive replacement indicator

Required

A hot-swappable drive must have a local indicator that shows which drive or drives are ready for replacement, facilitating the servicing process and improving reliability by reducing possible errors.

This indicator should be on the drive chassis, not on the screen. The Device Bay “eject” signal can be used to activate a replacement indicator. Designers can choose to use existing LEDs for dual purposes to fulfill this requirement, but the LED display should clearly show when a drive is ready for removal, as opposed to other information the display would normally provide.

For systems with multiple drives, an individual replacement indicator should be physically associated with each hot-swappable drive slot.

System includes alert indicators for imminence of failure

Alert indicators that indicate informative failure are required for Enterprise servers and should be provided for all servers. The hard failure and informative failure indicators cannot be on simultaneously. In addition to visual alerting mechanisms, a design can also provide software alerts such as paging, fax, or e-mail notifications.

The following are required sources of alert indicators for imminent failures for systems running Windows NT Server/Enterprise Edition. These are recommended for servers running other versions of Windows NT Server:

• Abnormal temperature of processor or inside chassis
  
  
• AC power line failure (operated by UPS)