The source Datareplicator executes the following processing for each transmission target:

• The source Datareplicator uses extraction information queue I/O buffers for transmission to read update information that is subject to transmission from the extraction information queue file.
• Of all the update information read from the extraction information queue file, the source Datareplicator extracts and edits only the update information for completed transactions and sends this information to the target system. This is the use the source Datareplicator makes of the update information editing buffer.

Additionally, the source Datareplicator uses the transaction management information buffer, which contains transaction management information, to manage the transaction completion status. The following figure provides an overview of the transmission procedure.

Figure 4-32 Overview of the transmission procedure

(a) Transaction management information buffer

When the source Datareplicator starts, it allocates a transaction management information buffer in local memory. This section discusses how you estimate and specify the size of the transaction management information buffer.

Estimating the size of the transaction management information buffer

• The transaction management information buffer is used to store the transaction management information contained in the extraction information queue file. The source Datareplicator uses this transaction management information buffer for the purpose of editing and sending update information. If the size of the transaction management information becomes greater than the buffer size during data extraction or transmission, the source Datareplicator adds an area whose size is one-fifth of the initial value. Such buffer addition might increase the system workload and slow down the processing speed. As the buffer size increases, the probability of buffer addition decreases. However, if the buffer size is too large, a memory shortage might occur. Therefore, carefully estimate an appropriate size for the transaction management information buffer so that neither area additions nor memory shortages occur.
• If the source HiRDB is a parallel server, you must estimate a transaction management information buffer size for each back-end server.

Specifying the size of the transaction management information buffer

• You use the maxtran and maxtrandata operands in the transmission environment definition to specify the size of the transaction management information buffer. The source Datareplicator multiplies the value of the maxtran operand, which is the maximum number of concurrently executable transactions, by the value of the maxtrandata operand, which is the maximum number of update information items in a transaction, and uses the resulting value for the initial size of the transaction management information buffer.

(b) Extraction information queue I/O buffers for transmission

When the source Datareplicator starts, it allocates extraction information queue I/O buffers for transmission in local memory. This section discusses how you estimate and specify the size and number of extraction information queue I/O buffers for transmission.

Estimating the size and number of extraction information queue I/O buffers for transmission

• The source Datareplicator uses extraction information queue I/O buffers for transmission to read update information stored in an extraction information queue file. It is necessary that the update information that is read be the same size as the stored update information that was extracted from the system log file. Therefore, an extraction information queue I/O buffer for transmission must be the same size as the extraction information queue I/O buffer for extraction. For details about the extraction information queue I/O buffer for extraction, see 4.6.3 Designing the extraction procedure.
• You can specify the number of extraction information queue I/O buffers for transmission. If you have specified a transmission interval, the source Datareplicator reads update information at the specified transmission interval, and then sends it. If there are many buffers, the source Datareplicator can use the update information in the buffers, thereby reducing the number of read operations on the extraction information queue file.

Specifying the size of an extraction information queue I/O buffer for transmission

• The size of an extraction information queue I/O buffer for transmission is the same as the size of the extraction information queue I/O buffer for extraction. For details about the extraction information queue I/O buffer for extraction, see 4.6.3 Designing the extraction procedure.

Specifying the number of extraction information queue I/O buffers for transmission

• You use the readbufnum operand in the transmission environment definition to specify the number of extraction information queue I/O buffers for transmission.

(c) Update information editing buffer

When the source Datareplicator starts, it allocates the update information editing buffer in local memory. This section discusses how you estimate and specify the size of the update information editing buffer.

Estimating the size of the update information editing buffer

• The source Datareplicator uses the update information editing buffer to edit update information. As the size of the update information editing buffer increases, more update information can be edited at one time, thereby reducing the number of read/write operations. However, if the update information editing buffer is too large, a memory shortage might occur.
• The update information editing buffer must be large enough to store at least one SQL update information item.
• If the source HiRDB is a parallel server, you must specify an update information editing buffer size for each back-end server.

Specifying the size of the update information editing buffer

• You use the editbufsize operand in the transmission environment definition to specify the size of the update information editing buffer.

The source Datareplicator sends the previous update information to the target in increments at a specified interval called the transmission interval. You use the sendintvl operand in the transmission environment definition to specify the transmission interval. The following figure provides an overview of the transmission interval for transmission processing.

Figure 4-33 Overview of transmission interval for transmission processing

If connection establishment with the target system fails, the source Datareplicator can retry the connection establishment; this is called connection retry. You use the retrynum operand in the transmission environment definition to specify the maximum number of times the source Datareplicator is permitted to retry connection establishment. If no connection retries count is specified, the source Datareplicator waits for the next transmission interval and attempts to establish connection until connection establishment is successful.

The source Datareplicator can send update information from one extraction environment to multiple destinations. However, if update information remains for some reason in the next extraction information queue file to be used, Datareplicator cannot swap extraction information queue files, making it impossible to extract the remaining update information from the system log file.

If the source Datareplicator cannot swap extraction information queue files, it cancels data linkage with the specified destination only, discards the corresponding update information from the next extraction information queue file to be used, and then swaps files. This is called reduced-mode transmission processing. You use the overwrite operand in the transmission environment definition to specify use of reduced-mode transmission processing.

The following figure shows the concept of reduced-mode transmission processing.

Figure 4-34 Concept of reduced-mode transmission processing

(a) Operation when reduced-mode transmission processing is specified

If update information remains in the next extraction information queue file to be used while Datareplicator cannot swap extraction information queue files, and reduced-mode transmission processing is specified for that update information's destination, Datareplicator stops only the corresponding transmission process and discards that update information only from the next extraction information queue file to be used. As a result, Datareplicator can then swap extraction information queue files and resume extracting update information from the system log file.

When reduced-mode transmission processing occurs, Datareplicator does not guarantee conformity between the source database and the target database at the destination that is subject to reduced-mode transmission processing. Therefore, you might have to re-create the target database at the destination.

If data linkage throughout the entire system is more important than conformity between the source database and the target databases at all destinations, specify reduced-mode transmission processing.

(b) Operation when reduced-mode transmission processing is not specified

When Datareplicator cannot swap extraction information queue files, it stops extracting update information from the system log file until space becomes available in the next extraction information queue file to be used; space will become available when the update information the file contains has finally been transmitted to its target system. Therefore, conformity is guaranteed between the source and target databases.

If it is important to guarantee conformity between the source and target databases in the overall data linkage system, do not specify reduced-mode transmission processing.

When the source HiRDB is a parallel server, the source Datareplicator requires a transmission environment definition for each back-end server. In such a case, the source Datareplicator's transmission environment definition can include the environment common to all back-end servers and the environment for a specific back-end server.

You define the common environment under commondef and a specific environment under besdef(server-name) in the transmission environment definition. If you specify the same operand in both commondef and besdef(server-name), the specification in besdef is effective for the back-end server specified in besdef.