Duplexing log data on a coupling facility log stream

System logger allows the installation to choose between different methods of duplexing coupling facility log data for a structure-based log stream:

Maintain a copy of coupling facility resident log data in local storage buffers on each system.
This option is the default, and occurs automatically if you do not specifically request duplexing on staging data sets in the log stream definition in the LOGR policy.
Maintain a copy of coupling facility resident log data in DASD staging data sets, one per log stream on a system.
This option provides hardening of the data on failure independent, persistent media.
Maintain a copy of coupling facility resident log data in a structure eligible for system-managed duplexing rebuild.
This option allows the coupling facility structure to be duplexed automatically, meaning that there will be two instances of the structure, one being the copy of the other. By using system-managed duplexing, a copy of the structure log data can be on a coupling facility that is separate from the primary instance of the structure. This provides the capability of having a backup of the structure log data at an “off-site” location (within the bounds of the CF-to-CF links).

The duplexing method for a coupling facility log stream is established on the log stream definition in the LOGR policy along with the CFRM policy structure definitions. See Selecting a method of duplexing coupling facility log data for details on how to specify the duplexing options that will be used to backup the log stream data in the structure.

As a default, system logger will provide its own log data duplexing. However, the logger duplexing can be used in conjunction with system-managed duplexing or it can be replaced by system-managed duplexing.

You can request the duplexing options for a coupling facility log stream in the log stream definition in the LOGR policy.

You can request duplexing to staging data sets either conditionally, so that a connection will have a staging data set when it contains a single point of failure, or unconditionally, so that every connection to a log stream duplexes to staging data sets. This means that the number of staging data sets active at any given time will depend on the status of each connection and the duplexing method selected for each log stream.

The log data duplexing to the staging data set can be done in a synchronous manner or asynchronous manner. Normally the log data will be duplexed synchronously, meaning that after it is written to the primary log stream storage, the log data is then written to the staging data set before Logger returning to the log writer with the results of the write request. However, when specifically requested, the log data can be duplexed asynchronously, meaning that after it is written to the primary log stream storage, the log data is then buffered in an I/O buffer but may be written to the staging data set after Logger returns to the log writer with the results of the write request.

If you request duplexing using staging data sets for a coupling facility log stream in the LOGR policy, system logger uses VSAM linear DASD staging data sets to hold a back-up copy of coupling facility log data for log streams that specify it.

IBM® suggests that you plan for DASD staging data sets even if you do not request duplexing to staging data sets. If the primary LOGR couple data set for the sysplex is at OS/390® Release 3 or above, system logger will automatically allocate a staging data set for a log stream under specific error conditions where the log data exists only in a volatile configuration. System logger does this to help safeguard log data against data loss.