You can use the import transaction synchronization facility only when the source and target systems are both Datareplicator.

To use this facility, all of the following conditions must be satisfied:

• The source HiRDB's version is 07-03 or later and it is a HiRDB/Parallel Server.
• The source Datareplicator's version is 07-04 or later.
• The target Datareplicator's version is 07-04 or later.

If you are using any of the following OSs, the target HiRDB's version must be 07-03 or later:

• HP-UX (IPF)
• Linux (IPF)
• Windows

If the system uses HiRDB Datareplicator Extension, data integrity cannot be guaranteed for a global transaction because HiRDB Datareplicator Extension does not support the XA interface.

(a) Correspondence between the source system's back-end servers and the target system's data linkage identifiers

To use this facility, you must provide a data linkage identifier for each back-end server of the source HiRDB. In this case, the data linkage identifiers must be provided as a synchronous import group. You must also place an event control table (hde_dtbl) at the source back-end servers.

The following figure shows an example of a configuration when the import transaction synchronization facility is used.

Figure 3-40 Example of a configuration when the import transaction synchronization facility is used

A global transaction that is extracted from a single source HiRDB must be received by a single target system. Therefore, the maximum number of back-end servers where a table subject to data linkage is placed is equal to the maximum number of global transactions that can be received by a single target Datareplicator (if the target system is UNIX, the maximum is 128; if it is Windows, the maximum is 63).

A back-end server that is not subject to extraction is not a target of a global transaction. As shown in the figure below, if a synchronous import group includes a data linkage identifier that corresponds to a back-end server that is not subject to extraction, import processing does not progress because the system waits for synchronization.

Figure 3-41 Example of configuration where the system waits for synchronization

(b) Configuration of source and target systems

When this facility is used, a single target system can receive only one global transaction. A single target system can receive both global and non-global transactions. The following figure shows an example of a configuration.

Figure 3-42 Example of a configuration of source and target systems (1)

Description

This example sends to a single target system a global transaction from source system 1 and a non-global transaction from source system 2.

A single global transaction can also be sent to multiple target identifiers. You can specify whether a global transaction is to be sent for each target identifier. The following figure shows an example of a configuration.

Figure 3-43 Example of a configuration of source and target systems (2)

Description

This example sends global and non-global transactions from a single source system to multiple target systems.

As shown in the figure below, a set of target identifiers cannot be distributed to multiple target systems. With this configuration, synchronization is not guaranteed.

Figure 3-44 Example of a configuration that cannot be synchronized

The following table shows the operands that must be specified in order to use the import transaction synchronization facility.

Table 3-19 Operands to be specified in order to use the import transaction synchronization facility

PP	Definition file	Operand	Description
HiRDB	System common definition	pd_rpl_reflect_mode = uap#1	Outputs transaction branch information (needed for Datareplicator to achieve this facility) to the system log file. For details, see the manual HiRDB Version 9 System Definition.
Datareplicator	Import system definition	syncgroup001#2	Constitutes a synchronous import group.
	Transmission environment definition	reflect_mode = uap#3	Sends transaction branch information to the target Datareplicator.
		eventsync	Specifies the number of the synchronous event (that triggers issuance of COMMIT by the synchronous import group).

#1

If you change the value from uap to server or vice versa during data linkage, the KFRB03312-E message might be output during import processing, resulting in termination of import processing. In such a case, transmission of the transaction branch information stops. To avoid this, you must release the synchronous import group, and then restart.

#2

If transaction branch information is to be sent but the syncgroup001 operand is not defined, the KFRB02066-W message is output and import processing is executed asynchronously. If transaction branch information is not to be sent but the syncgroup001 operand is defined, a port check error (default code = 26) results.

#3

If HiRDB's pd_rpl_reflect_mode operand value is server, an error occurs during execution of the hdestart command and the KFRB00726-E message is output even when uap has been specified in the reflect_mode operand.

If you specify uap in the reflect_mode operand, specify false in the overwrite operand in the transmission environment definition. If the specified value is not false, the KFRB00846-E message is output, resulting in an error.