Performance Manager Commands › ADVISE PERFORMANCE EXPORT › Change a Rule Literal Value › Qualifiers

Qualifiers

/BEGINNING=time

Specifies the starting date and time of the data to be exported. If /ENDING is not specified, the default /BEGINNING time is TODAY. If /ENDING is specified, the default /BEGINNING time is 00:00 of the date specified with /ENDING.

/BEGINNING is incompatible with the /DATES qualifier.

/CLASS=(item[,...])

Specifies which optional classes of statistics are included in the export file.

You can select any of the following class items:

You can negate these keywords to indicate that reporting of a particular class of data is not wanted. Also, you can specify /CLASS=(ALL[,negated-keyword]) to allow an “all but these” capability.

The DEFAULT_STATISTICS class of statistics is always on unless specifically negated. For example, if /CLASS=LOCK is specified, then CPU, DISK, IO, MEMORY, PAGE_FAULT, XQP_CACHE and LOCK statistics are exported. To disable any classes of data which are part of the DEFAULT_STATISTICS group, you must specify NODEFAULT_STATISTICS. Therefore, if you want to report only disk and CPU data, specify /CLASS=(NODEFAULT_STATISTICS,CPU,DISK).

The PROCESS keyword allows you to specify the optional keywords EXTENDED, IMAGE or ALL. The presence of the EXTENDED keyword indicates that the extended process metric data is to be exported along with the standard process metric data. If you do not specify EXTENDED, then NOEXTENDED is assumed. The presence of the IMAGE keyword indicates that the image name for each process is to be exported along with the standard process metric data. If you do not specify IMAGE, then NOIMAGE is assumed. The presence of the ALL keyword indicates that all available process metric data is to be exported, including the standard process metric data, the extended process metric data, and the image name.

If you omit the /CLASS keyword from the command line, then /CLASS=DEFAULT_STATISTICS is assumed.

/CLASSIFY_BY=USERGROUP=workload-family

Lets you specify a workload family to control how process data is classified. By default, all process data is exported without being summarized.

/COLLECTION_DEFINITION=collection-definition-name

Specifies the name of the Collection Definition, and hence the collected data that you want to export. If you omit this qualifier, daily data is obtained from the Collection Definition called “CPD.” To view the Collection Definitions that you have available, use the DCL command ADVISE COLLECT SHOW SCHEDULE. If you want to use history data instead of daily data, use the /HISTORY qualifier instead of the /COLLECTION_DEFINITION qualifier. These two qualifiers are mutually exclusive.

/DATES=filespec

Specifies that a file containing a series of date ranges is to be used in place of the /BEGINNING and /ENDING qualifiers. Each line in the dates file should look as follows:

dd-mmm-yyyy hh:mm:ss.cc,dd-mmm-yyyy hh:mm:ss.cc

The time can be either omitted entirely or truncated. Any truncated parts of the time default to 0. The periods of time represented by each line in the file need not be contiguous but they must be in ascending order.

/DATES is incompatible with the /BEGINNING and /ENDING qualifiers.

/ENDING=time

Specifies the ending date and time of the data to be exported. If you do not specify /BEGINNING, the default ending time is the current time. If you specify /BEGINNING, the default ending time is midnight (23:59) for the same day.

/FILTER=keyword

The /FILTER qualifier allows you select a subset of the daily or history data for exporting. Process data and disk data can be filtered.

Process data can be filtered by using any of the filter keywords: USERNAMES, IMAGENAMES, PROCESSNAMES, ACCOUNTNAMES, UICS, PIDS or WORKLOADNAMES. If a process record's identification information matches any of the identification specifications that are specified, then that record is selected.

Likewise, disk data can be filtered by using either of the filter keywords, VOLUMENAMES and DEVICENAMES. If a device record's identification information matches any of the volume names or device names that are specified, then that record is selected.

The following table lists the keywords:

/HISTORY=history-descriptor-name

Allows you to select history data from the Performance Manager database. By default, daily .CPD files are processed. However, by specifying the name of a history file descriptor you can select historical data instead.

You must have previously defined the descriptor-name in the parameters file and have used the archiving facility to create the history files. Use the DCL command ADVISE EDIT to start the parameters editor. From the utility, you can ADD, DELETE, MODIFY, and SHOW history file descriptors.

If you want to use history data instead of daily data, use the /HISTORY qualifier instead of the /COLLECTION_DEFINITION qualifier. These two qualifiers are mutually exclusive.

/INTERVAL=seconds

Specifies the elapsed time to be summarized in an output record. Its minimum value is that of the performance data file being exported. It must be a multiple of the interval of the data file, and are rounded up to match such a value.

The default value is the interval of the data file being exported.

The value is expressed in seconds.

/NODE_NAME=node-name

Identifies the node for which data is to be exported. /NODE_NAME is required if your collection definition supports multiple nodes. Only one node's data can be written out to an export file.

/OUTPUT[=file-spec]

Specifies the name of the export file. If you do not specify /OUTPUT, or if you specify /OUTPUT without a file specification, a default filename of PSPA$DUMP.DAT is used.

/PROCESS_STATISTICS=([PRIMARY_KEY=option] [,SECONDARY_KEY=option])

This qualifier allows you to specify how process records can be summarized and sorted, and to indicate what output fields to preserve in the output. By default, all details are preserved, and all fields are supplied. The default settings are:

/PROCESS_STATISTICS=(PRIMARY_KEY=PID,

SECONDARY_KEY=IMAGENAME)

If you use /CLASSIFY_BY to attach workload information to the process records, you affect the default PRIMARY_KEY, which then becomes WORKLOAD_NAME. To obtain all detail fields (such as PID), including the workload name, you must specify both the /CLASSIFY_BY qualifier, and the /PROCESS_STATISTICS = (PRIMARY_KEY=PID, SECONDARY_KEY=IMAGENAME) settings.

Primary key options are:

PRIMARY_KEY={MODE|USERNAME|IMAGENAME| UIC_GROUP|PROCESS_NAME| WORKLOAD_NAME|ACCOUNT_NAME|PID}

Secondary key options are:

SECONDARY_KEY={MODE|USERNAME|IMAGENAME| UIC_GROUP|PROCESS_NAME| WORKLOAD_NAME|ACCOUNT_NAME|PID}

If the value for secondary key is the same as for the primary key, no secondary level breakout occurs. This also happens if the primary key is specified and no secondary key is given.

The following example demonstrates the use of this qualifier:

$ ADVISE PERFORMANCE EXPORT /CLASS=(NODEFAULT, PROCESS)/NODE=MYNODE - _$ /BEGIN=time/END=time -
_$ /PROCESS_STATISTICS=(PRIMARY_KEY=IMAGENAME,SECONDARY_KEY=PID) -
_$ /CLASSIFY_BY=USERGROUP=MODEL_TRANSACTIONS

This command produces an output file called PSPA$DUMP.DAT in the current directory, in ASCII CSV format. The process data records contain all detail fields, including the workload name with which it is associated, sorted by imagename first, then PID within that. The /CLASSIFY_BY qualifier is needed to specify the Family definition (and Workload definitions) that define the workload groups.

/SCHEDULE=({day=(hour-range[,...]) | NOday}[,...])

Where day is one of the following words:

• EVERYDAY
• WEEKDAYS
• WEEKENDS
• MONDAY
• TUESDAY
• WEDNESDAY
• THURSDAY
• FRIDAY
• SATURDAY
• SUNDAY

The hour-range is in the form of m-n where m is an integer (hour) from 0 to 23 and n is an integer (hour) from 1 to 24 and larger than m.

Use the /SCHEDULE qualifier to select a sub-set of the performance data for exporting. By default all data between the /BEGINNING time and the /ENDING time is selected. Use the day keywords with hour ranges to specify what data is to be included.

Negate any of the day keywords to omit data for a range of days. Do not specify any hour ranges with negated keywords.

/TYPE={ASCII | BINARY}

There are two different data types available for the exported file, ASCII and BINARY.

The default type is ASCII which provides data in CSV (comma separated variable) format with text items in quotes. For examples commands and the resulting output data, see Appendix E.

To use either format in subsequent processing of the data, two files have been placed in the PSPA$EXAMPLES directory area, PSPA$DUMP_ASCII.DTR and PSPA$DUMP_BINARY.DTR. To specify a format, supply a value to the qualifier of “ASCII” or “BINARY.” The default is “BINARY.”