By specifying timestamp in the to clause of the import table definition, you can obtain a historical record of update information in chronological order. However, time-ordered information might not be available, depending on the type of update processing. If Datareplicator cannot obtain the specified time-ordered information, it sets the null value.

The following table shows the contents and availability of time-ordered information.

Table 4-30 Contents and availability of time-ordered information

Information	Description	Availability
		Update	Insert	Delete	PURGE
Extraction date	Date update information was output to the update journal at the source system	Y	Y	Y	Y
Extraction time	Time update information was output to the update journal at the source system	Y	Y	Y	Y
Extraction time stamp	Date and time update information was output to the update journal at the source system	Y	Y	Y	Y
Import date	Date update information was imported into the target database at the target system	Y	Y	Y	Y
Import time	Time update information was imported into the target database at the target system	Y	Y	Y	Y
Import time stamp	Date and time update information was imported to the target database at the target system	Y	Y	Y	Y
Import type	Import type, indicated as follows: Update: upd Insert: ins Delete: del PURGE: purge	Y	Y	Y	Y
Mapping key	Key value used to identify the data to be updated	Y	Y	Y	N/A
Update information	Updated data	Y	Y	N/A	N/A

Y: Available.

N/A: Becomes the null value.

To collect time-ordered information, such as update information, import dates, and import times, you must create in advance in the target HiRDB database a table for storing this information (time-ordered information table). You can specify any name for the table and you must specify the table's column structure (number of columns, order of columns, column names). You must define the column attributes as appropriate for the information to be collected. Depending on the type of information, you might have to specify a literal in the const clause in the update information field definition.

The following table shows the column attributes of a time-ordered information table and the specification in the const clause.

Table 4-31 Column attributes for a time-ordered information table and the specification in the const clause

Information to be collected	Column attribute	Specification in the const clause in the update information field definition
Extraction date	DATE	extract_date
Extraction time	TIME	extract_time
Extraction time stamp	TIMESTAMP[(p)]#, p = 0, 2, 4, 6	extract_timestamp
Import date	DATE	reflect_date
Import time	TIME	reflect_time
Import time stamp	TIMESTAMP[(p)]#, p = 0, 2, 4, 6	reflect_timestamp
Import type	CHAR(5)	reflect_kind
Mapping key	Attribute that allows storage of mapping keys	N/A
Update information	Attribute that allows storage of update information	N/A

N/A: Not applicable.

#

p is an integer representing the number of decimal places displayed for the seconds value. For example, p = 2 means that the seconds value is to be obtained out to two decimal places.

Note the following about using time-ordered information:

• You use HiRDB's retrieval facility (SELECT command) to reference time-ordered information. However, the information might not be collected in the order in which the source database was updated. To retrieve the information in the source database's updating order, you must sort it during retrieval.
• Information, such as update information, import dates, and import times, keeps accumulating over time in a time-ordered information table. Therefore, you need to periodically delete unneeded information from this table by issuing the DELETE or PURGE TABLE statement.

• If data obtained from BLOB column partitions has been imported to a time-ordered table, only the data obtained from the BLOB column partitions can be imported.
• Do not import to a time-ordered information table BLOB-type or BINARY-type data to which backward deletion updating has been performed. If such data is imported to a time-ordered information table, inconsistency occurs between the source and target databases.