21.10 Tuning the system's internal processing

In some cases, you can specify an operand to facilitate the system's internal processing.

These operands are described in this section. For details about each operand, see the section on each operand in the manual HiRDB Version 8 System Definition.

Organization of this section
(1) Operands for increasing performance, even if the machine load increases
(2) Operands for specifying performance during concurrent transaction execution

(1) Operands for increasing performance, even if the machine load increases

The operands described here are used to select among methods that use one of the following two general approaches:

  1. An approach that places a high load on the machine, but provides high performance (for a machine capable of high performance). If the machine's performance capacity is low, CPU usage may approach 100%, posing the risk of degraded OS performance.
  2. An approach that places a low workload on the machine, but can ensure standard performance. Stable performance can be achieved with any type of machine.

To minimize risk, the default values for many of these operands are set to use a method that adheres to approach 2.

If your machine's performance capacity is high, you may be able to improve performance (while also increasing the load on the machine) by changing the values specified for these operands. However, if unsuitable values are specified, the CPU usage may approach 100%, posing the risk of degraded OS performance. Thus, you should be careful about changing the specification values for these operands.

Note that these operands are not described in detail here because their specification values are related to the system's complex internal processing. You should refer to the specification guidelines for each operand that are provided in the manual HiRDB Version 8 System Definition.

Table 21-2 lists the operands for tuning the system's internal processing.

Table 21-2 Operands for tuning the system's internal processing

Specification itemSpecification conditionOperand
Lock-release detection methodFor general UAP executionpd_lck_release_detect
Lock-release detection intervalpd_lck_release_detect_interval
Processing request queuing methodpd_server_entry_queue
Inter-thread lock sleep methodpd_thdlock_sleep_func
Inter-thread lock sleep timepd_thdlock_retry_time
Inter-thread spin lock spin countpd_thdspnlk_spn_count
Global buffer lock-release detection intervalGlobal buffer lock-release-wait has occurred.pd_dbbuff_wait_interval
Spin count during global buffer lock release detectionpd_dbbuff_wait_spn_count
Global buffer lock-release detection methodGlobal buffer is being used.pd_dbbuff_lock_release_detect
Spin count during global buffer lock-acquisition-waitpd_dbbuff_lock_spn_count
Interval during global buffer lock-acquisition-waitpd_dbbuff_lock_interval

(2) Operands for specifying performance during concurrent transaction execution

Table 21-3 shows the operands for specifying performance during concurrent transaction execution.

Table 21-3 Operands for specifying performance during concurrent transaction execution

Specification itemSpecification conditionOperand
Inter-thread lock release notification methodWhen the execution time must be the same for all transactions that are being executed concurrently.pd_thdlock_wakeup_lock
Inter-thread lock release detection intervalWhen the execution times may differ to some extent among the transactions that are being executed concurrently.pd_thdlock_pipe_retry_interval