33.2.2 Format of merge chunk options

(1) Specification format

[set adb_mergechunk_rthd_num = number-of-processing-real-threads-to-be-used-by-adbmergechunk-command]
[set adb_mergechunk_scan_buff_size = table-data-scan-buffer-size]
[set adb_mergechunk_buff_blk_num = number-of-buffer-blocks-for-merge-chunks]
[set adb_mergechunk_sort_buff_size = sort-buffer-size]
[set adb_mergechunk_dvix_rdbuff_size = B-tree-and-text-index-input-buffer-size]
[set adb_mergechunk_dvix_wtbuff_size = B-tree-and-text-index-output-buffer-size]
[set adb_mergechunk_txt_buff_size = text-index-creation-buffer-size]

The specification rules for these merge chunk options are the same as for the server definition. For details about the syntax rules for the server definition, see the topic Syntax rules for the server definition in the HADB Setup and Operation Guide.

To Page Top

(2) Explanation of specification format

• adb_mergechunk_rthd_num = number-of-processing-real-threads-to-be-used-by-adbmergechunk-command

~<integer>((3 to 4,096))

Specifies the number of processing real threads to be used for executing the adbmergechunk command.

Specification of this option is recommended. The adbmergechunk command uses the number of processing real threads specified by this option. Therefore, specify an appropriate number of processing real threads for use by the adbmergechunk command. For details about the processing of the adbmergechunk command that uses processing real threads and the number of processing real threads to be used for each step of processing, see 33.1.2 Relationships among threads used by the adbmergechunk command.

If this option is omitted, the number of processing real threads to be used for executing the adbmergechunk command is determined by the adb_sql_exe_max_rthd_num operand in the server definition. The following table shows the relationship between the value of this option or the adb_sql_exe_max_rthd_num operand in the server definition and the number of processing real threads used for executing the command.

Table 33‒3: Number of processing real threads used for executing the command

No.	Value in this option or in the server definition		Number of processing real threads used for executing the command
1	When this option is omitted	If the adb_sql_exe_max_rthd_num operand in the server definition is not specified	The default value of the adb_sql_exe_max_rthd_num operand in the server definition is used.
2		If a value in the range from 0 to 2 is specified in the adb_sql_exe_max_rthd_num operand in the server definition	--
3		If 3 or a greater value is specified for the adb_sql_exe_max_rthd_num operand in the server definition	The value specified for the adb_sql_exe_max_rthd_num operand in the server definition is used.
4	If 3 or a greater value is specified in this option		The value specified in this option is used.

Legend:

--: The adbmergechunk command results in an error.

Note:

For details about the adb_sql_exe_max_rthd_num operand in the server definition, see Operands related to performance (set format) in the HADB Setup and Operation Guide.

• adb_mergechunk_scan_buff_size = table-data-scan-buffer-size

~<integer>((1 to 2,048))<<8>> (megabytes)

Specifies the size (in megabytes) of the buffer to be used for retaining table data retrieval results. A larger value can reduce overhead in retrieving table data.

As a guideline, specify the value obtained from the following formula.

Formula

Key length of the B-tree indexes defined for the table to be processed × 100,000

For details about how to determine the key length of a B-tree index, see Determining the key length (KEYSZ) of a B-tree index in Estimating the size of the data DB area in Designing a Database in the HADB Setup and Operation Guide.

• adb_mergechunk_buff_blk_num = number-of-buffer-blocks-for-merge-chunks

~<integer>((5 to 4,096))<<64>>

Specifies the number of buffer blocks to be used for merge chunk processing. You can use the default value, but a larger value might improve the speed of merge chunk processing.

• adb_mergechunk_sort_buff_size = sort-buffer-size

~<integer>((1 to 2,048))<<256>> (megabytes)

Consider specifying this option when B-tree or text indexes are defined for the table to be processed.

Specifies the size (in megabytes) of the sort buffer to be used for creating B-tree and text indexes. Specify a large value if there is sufficient memory. Better performance can be expected as this option's value becomes larger.

• adb_mergechunk_dvix_rdbuff_size = B-tree-and-text-index-input-buffer-size

~<integer>((32 to 51,200))<<1,024>> (kilobytes)

Consider specifying this option if B-tree or text indexes are defined for the table to be processed.

Specifies the size (in kilobytes) of the input buffer to be used for creating B-tree and text indexes. Specify a large value if there is sufficient memory. Better performance can be expected as this option's value becomes larger.

• adb_mergechunk_dvix_wtbuff_size = B-tree-and-text-index-output-buffer-size

~<integer>((32 to 51,200))<<1,024>> (kilobytes)

Consider specifying this option if B-tree or text indexes are defined for the table to be processed.

Specifies the size (in kilobytes) of the output buffer to be used for creating B-tree and text indexes. Specify a large value if there is sufficient memory. Better performance can be expected as this option's value becomes larger.

• adb_mergechunk_txt_buff_size = text-index-creation-buffer-size

~<integer>((1 to 2,048))<<256>> (megabytes)

Consider specifying this option if text indexes are defined for the table to be processed.

Specifies the size (in megabytes) of the buffer to be used for creating text indexes.

Increasing the value specified for this option often improves performance. However, the amount of buffer memory specified by this option will be used for all processing real threads on which text indexes will be built. Therefore, be careful when you increase the value specified for this option.

To Page Top