NonStop Systems Introduction for H-Series RVUs

Integrity NonStop NS-Series Server Architecture
NonStop Systems Introduction for H-Series RVUs—540083-001
7-5
Fault Tolerance of Integrity NonStop Servers
Fault Tolerance of Integrity NonStop Servers
The hardware, operating system, and application environment of Integrity NonStop
servers work together to meet all the demanding requirements of a real-time
enterprise, such as fault tolerance, system expandability, and high performance.
None of these requirements is more important than fault tolerance, and the hardware
architecture is especially designed to achieve fault tolerance for applications.
Significantly, the hardware is designed to provide two forms of continuous availability
when an individual component fails:
•
Continuous execution of processes
•
Continued access to databases
Figure 7-4 illustrates both of these design goals.
The system achieves fault tolerance for user processes and system processes by
running programs as process pairs in different processors. The primary process
executes actively in one processor while the backup process remains in a wait state in
another processor. At critical points, the primary process sends checkpoint messages
over the dual ServerNet fabrics to the backup process.
If the processor that is executing the primary process fails, the backup process in the
healthy processor takes over where the primary process left off. It resumes execution
of the work from the last valid checkpoint and accesses the same database on disk
that the primary process had been using.
Figure 7-4. Continuous Availability of an Integrity NonStop Server
Processor 0
Processor 1
Checkpoint
messages
Primary
process
Backup
process
Mirrored
disk volumes
VST092.vsd