14.1 Overview

Executor: HiRDB administrator

The statistics analysis utility reads information input from statistics unload files and system log files and edits statistical information.

The statistics analysis utility enables needed statistical information to be edited and output by specifying editing beginning and ending points. The user can obtain the activity status of the HiRDB system by analyzing this statistical information. The utility also outputs the statistical information editing results to the standard output, and it outputs the unedited statistical information to a DAT-format file. The user can create desired statistical reports from the DAT-format file.

The figure below provides an overview of the statistics analysis utility (pdstedit).

Figure 14-1 Overview of the statistics analysis utility (pdstedit)

[Figure]

Organization of this section
(1) Base files for statistical information
(2) Types of statistical information
(3) Execution conditions of the statistics analysis utility

(1) Base files for statistical information

The statistics analysis utility uses the following three files as its input files:

(2) Types of statistical information

The table below describes the types of statistical information that the statistics analysis utility capable of editing. You collect statistical information (except CONNECT and DISCONNECT information) by specifying the pdstbegin operand in the system definition or by executing the pdstbegin command. To collect statistical CONNECT and DISCONNECT information, you specify the pdhibegin -k cnc operand in the system definition.

Table 14-1 Overview of statistical information output by statistics analysis utility

Type of statistical informationOverview
System activity statistical informationEdits and outputs activity information for processes, RPC, and logs in HiRDB systems by HiRDB system or by server.
UAP statistical information#1Edits by editing interval information on UAPs, such as the number of selection rows and the number of times each SQL statement is executed, and outputs the information by UAP or service.
SQL statistical information#1Edits and outputs by UAP or service information about SQL statements issued, such as the number of selection rows, the number of times a work table was created, and the number of back-end servers that issued an SQL split command.
Global buffer pool statistical information#2Edits by editing interval information about global buffer accesses, such as the buffer hits rate and the number of real I/O operations, and outputs the information by server or global buffer.
Statistical information on HiRDB files for database manipulation#2Edits by editing interval information on HiRDB file accesses, such as the number of synchronous I/O operations and the number of I/O errors, and outputs the information by server, HiRDB file, or RDAREA.
Deferred write processing statistical informationEdits by editing interval information on deferred write processing, such as the number of deferred write operations, the cause of each operation, and the concurrency level of I/O operations, and outputs the information by server.
Index statistical information#3Edits by editing interval information on indexes, such as index key lock information and index split information, for the statistics log and the system log, and outputs the information by server or index. This information cannot be output to a DAT-format file.
SQL static optimization informationOutputs SQL static optimization information. This information can be output to a DAT-format file only.
SQL dynamic optimization informationOutputs SQL dynamic optimization information. This information can be output to a DAT-format file only.
SQL object execution informationOutputs SQL object execution information. This information can be output to a DAT-format file only
SQL statement statistical informationOutputs issued data manipulation SQL, definition SQL, and LOCK statements and SQL information. This information can be output to a DAT-format file only.
CONNECT/DISCONNECT statistical information#4Outputs CONNECT and DISCONNECT information. This information can be output to a DAT-format file only.
SQL object transfer statistical informationOutputs SQL object transfer information. This information can be output to a DAT-format file only.
Statistical information about database I/O operationsOutputs statistical information about database I/O operations. This information can be output to a DAT-format file only.

#1: UAP and SQL statistical information may not agree, due to differences in the timing of entry of the pdstbegin and pdstend commands. Also, the number of SQL statements actually issued by the UAP and the number displayed may not agree.

#2: When a synchronization point is not generated within the time specified by the -t option or when only one synchronization point is generated, information for that interval of time is not output. A longer time interval should be specified in the -t option in this case.

#3: Index split information, which is one of the index statistical information items, is collected from the unload log files or system log files. To edit index split information, the file groups of the unload log files or the system log files must be used as the input. Other information is obtained from unload statistics log files.

#4: CONNECT and DISCONNECT statistical information is collected from unload log files or system log file groups. Therefore, use unload log files or system log file groups as the input to the statistics analysis utility. If the pdhibegin -k cnc operand is specified in the system definition, this statistical information is output to the single-server or to a server with a front-end server.

(3) Execution conditions of the statistics analysis utility

  1. You can execute the statistics analysis utility whether or not HiRDB is active.
  2. You can execute the statistics analysis utility on any server machine.