When a system switchover occurs, the following processing is performed to start the standby HiRDB:

• System server startup processing
• System file transfer processing
• Server process startup processing
• Roll forward processing

Of the preceding, the time needed for the server process startup processing accounts for a large proportion of the entire system switchover time. The time required for server process startup processing is directly proportional to the number of resident server processes, which means that system switchover time increases as the number of resident processes increases. You can reduce this time, however, by starting the server processes of the standby HiRDB beforehand, so that server process startup processing is not performed when the system is switched over. This allows you to reduce the system switchover time by the amount of time required for system process startup processing. This concept is called user server hot standby. For example, on a server machine that runs at a speed of approximately 100 MIPS, one second is required to start a single server process. In this case, user server hot standby would reduce the system switchover time by about one second for each resident server process.

For details about user server hot standby, see the HiRDB Version 8 System Operation Guide.

A facility is provided to start the server processes and the system server on the standby HiRDB beforehand, so that startup processing of the system processes and system server need not be performed when the system is switched over. This facility is called the rapid system switchover facility. With it, you can reduce the system switchover time by the amount of time required to perform startup processing of system processes and the server system. For details about the rapid system switchover facility, see the HiRDB Version 8 System Operation Guide.

Note that the rapid system switchover facility can reduce the system switchover time more than user server hot standby can (the rapid system switchover facility includes the user server hot standby function). Figure 8-11 compares the system switchover times of various operating modes.

Figure 8-11 Comparison of system switchover times

Explanation

The processing indicated in the shaded area is performed in advance, after which the associated processes stand by; therefore, this processing does not need to be performed when a system is switched over. Thus, system switchover time is reduced by the amount of time required to perform the processing indicated in the shaded area.