Availability Guide for Problem Management

Auditing Systems for Fault Tolerance
Availability Guide for Problem Management–125509
7-7
Using Mirrored Disk Drives
running at an average of 95 percent busy during peak periods. If a single processor were
to fail during such a peak period, it is highly unlikely that the remaining processors
would be able to perform well enough to take over the processing requirements of the
downed processor.
For more information about performance management, refer to the following:
•
Section 9, “Problem Management Tools”
•
Availability Guide for Performance Management
•
Introduction to NonStop Operations Management
Using Mirrored Disk Drives
One effective protection against loss of data is the use of mirrored disk volumes, which
maintain copies of data on two physically independent disk drives that are accessed as a
single device and are managed by the same I/O process. All data written to one disk is
also written to the other disk. All data read from one disk could be read from the other,
because the data is identical. A mirrored volume protects data against single-disk
failures: if one disk drive fails, the other remains operational. The probability of both
disk drives failing at the same time is low when you always have a disk drive repaired or
replaced promptly if it fails.
Avoiding a System Freeze
A freeze-enabled processor can assert a processor freeze on any other freeze-enabled
processor and can result in a system freeze. As a result, it is possible for one processor
failure to cause a system freeze. You can avoid a system freeze by ensuring that the
system-freeze flags are set correctly.
Note. If a single processor fails, you need to redistribute the load between the remaining
processors. With careful planning, you will be able to redistribute the load successfully.
However, you must consider how this redistribution will affect your applications. Typically, when
additional load is absorbed by remaining processors, application response time is affected.