HP 3000 Manuals HP 3000 Manuals
Overview of the Export Action (continued)
Data Items for Exporting Data (continued)
Table 5-5. EXPORT Items for Data Type APPLICATION (Multiple Layout)
Used with multiple application layouts. The following fields will be repeated once
for each application defined on the system.
Application Identification Metrics
@INTERVAL Time covered by this record (in seconds).
@SAMPLES Number of samples averaged into this record.
@APPLICATION_NO Sequential application number (matches PARM
file).
@APPLICATION Application name (20-byte ASCII).
@BLANK An empty field used as a spreadsheet place
holder.
Summary Metrics
@CPU_TOTAL CPU usage by the application during the
interval (percentage of total).
@CPU_SECONDS CPU usage by the application during the
interval (time, in seconds).
@DISC_TOTAL Physical disc IO rate (IOs/second).
@DISC_IO Number of physical disc IOs.
CPU Metrics
@CPU_LINEAR CPU usage while in the linear execution
queue (percentage of total).
@CPU_LINEAR_SECONDS CPU usage while in the linear execution
queue (time, in seconds).
@CPU_CS CPU usage while in the "CS" execution queue
(percentage of total).
@CPU_CS_SECONDS CPU usage while in the "CS" execution queue
(time, in seconds).
@CPU_DS CPU usage while in the "DS" execution queue
(percentage of total).
@CPU_DS_SECONDS CPU usage while in the "DS" execution queue
(time, in seconds)
@CPU_ES CPU usage while in the "ES" execution queue
(percentage of total).
@CPU_ES_SECONDS CPU usage while in the "ES" execution queue
(time, in seconds).
Disc Metrics
@DISC_LINEAR Physical disc IO rate while in linear
execution queue (IOs/second).
@DISC_LINEAR_IO Number of physical disc IOs while in the
linear execution queue.
@DISC_CS Physical disc IO rate while in the "CS"
execution queue (IOs/second).
@DISC_CS_IO Number of physical disc IOs while in the
"CS" execution queue.
@DISC_DS Physical disc IO rate while in the "DS"
execution queue (IOs/second).
Table 5-5. EXPORT Items for Data Type APPLICATION (Multiple Layout) (cont.)
@DISC_DS_IO Number of physical disc IOs while in the
"DS" execution queue.
@DISC_ES Physical disc IO rate while in the "ES"
execution queue (IOs/second).
@DISC_ES_IO Number of physical disc IOs while in the
"ES" execution queue.
Process Queue Depths
@CPUQUEUE Average number of processes waiting for or
using CPU.
@DISCQUEUE Average number of processes waiting for
DISC.
@MEMORYQUEUE Average number of processes waiting for
MEMORY.
@IMPEDEQUEUE Average number of processes waiting for
IMPEDES (locks).
Process Count Metrics
@AVE_PROCESSES Average number of processes in the
application.
@ACTIVE_PROCESSES Average number of processes that used CPU.
@PROCESSES_COMPLETED Number of application processes that
completed.
@PROCESSES_RUNTIME Average run time of completed processes (in
seconds).
Terminal Transaction Metrics
@TRANSACTIONS Number or rate of terminal transactions
completed during the interval.
@TRANSACTION_COUNT Number of terminal transactions completed
during the interval.
@THINKTIME Average transaction think time (in seconds).
@FIRSTRESP Average transaction time-to-first-response
(in seconds).
@PROMPT Average transaction response-to-prompt time
(in seconds).
@SERVICE_LEVEL Transactions that met service level
(percentage of total).
@SERVICE_LEVEL_COUNT Number of transactions that met service
level.
Average Process Wait State Metrics
@STOPCPU Time processes spent waiting for CPU
(percentage of total).
@STOPDISC Time processes spent waiting for DISC
(percentage of total).
@STOPIMPEDE Time processes spent IMPEDED (locked)
(percentage of total).
@STOPOTHERIO Time processes spent waiting for other IO
(percentage of total).
@STOPSWAP Time processes spent waiting for MEMORY
(percentage of total).
@STOPTERM Time processes spent waiting for terminal
input (percentage of total).
PROCESS Data Type.
The following data items are available for the PROCESS data type. There
is one record for each interesting process.
Table 5-6. EXPORT Items for Data Type PROCESS
Record Identification Metrics
RECORD_TYPE ASCII field to identify this record type
"PROC".
DATE Date in MM/DD/YY format (or custom NLS
date).
TIME Time in HH:MM 24-hour format.
DAY Julian day-of-the-year (1-366).
YEAR Year (such as 1991).
DATE_SECONDS Date in UN*X format (in seconds, since
January 1, 1970).
INTERVAL Time included in this sample (in seconds).
BLANK An empty field used as a spreadsheet place
holder.
Process Identification Metrics
PIN Process Identification Number.
PROGRAM Program name (or ":" last MPE command).
JSNO Job/Session number (J###/S### or "SYS").
LDEV Logon device number.
LOGON User logon string.
QUEUE Execution queue (L,A,B,C,D,E).
PRIORITY Execution priority at end of interval.
INTEREST Codes for reason of interest (see table
5-7).
STOP_REASON Reason process last stopped execution.
APPLICATION_NO Sequential application number (matches PARM
file).
Table 5-6. EXPORT Items for Data Type PROCESS (cont.)
Summary Metrics
RUNTIME Total process execution time until end of
interval.
TOTAL_CPU Percentage of CPU usage during process life
(percentage of total run time).
TOTAL_CPU_SECONDS Time of CPU usage during process life (run
time, in seconds).
TOTAL_DISC Physical disc IO rate during process life
(IOs/second).
TOTAL_DISC_IO Number of physical disc IOs during process
life.
TOTAL_FIRSTRESP Average transaction first-response time
during process life.
TOTAL_PROMPT Average transaction response-to-prompt time
during process life.
TOTAL_THINK Average transaction think time during
process life.
TOTAL_TRANS Number or Rate of transactions during
process life.
TOTAL_TRANS_COUNT Number of transactions during process life.
CPU Metrics
CPU CPU usage during the interval (percentage of
total).
CPU_SECONDS CPU usage during the interval (time, in
seconds).
CM CPU usage that was in compatibility mode
(percentage of total).
CM_SECONDS CPU usage that was in compatibility mode
(time, in seconds).
SWITCHTOCM Rate of switches from native mode to
compatibility mode (switches/second).
SWITCHTOCM_COUNT Number of switches from native mode to
compatibility mode.
SWITCHTONM Rate of switches from compatibility mode to
native mode (switches/second).
SWITCHTONM_COUNT Number of switches from compatibility mode
to native mode.
Disc metrics
DISC Physical disc IO rate (IOs/second).
DISC_IO Number of physical disc IOs.
LOGLREAD Logical disc read rate during interval
(IOs/Second).
LOGLREAD_IO Number of logical disc reads during the
interval.
LOGLWRITE Logical disc write rate during interval
(IOs/Second).
LOGLWRITE_IO Number of logical disc writes during the
interval.
Table 5-6. EXPORT Items for Data Type PROCESS (cont.)
PHYSREAD Physical disc read rate during interval
(IOs/Second).
PHYSREAD_IO Number of physical disc reads during the
interval.
PHYSWRITE Physical disc write rate during interval
(IOs/Second) .
PHYSWRITE_IO Number of physical disc writes during the
interval.
Terminal Transaction Metrics
TRANSACTIONS Number or rate of transactions completed
during interval.
TRANSACTION_COUNT Number of transactions completed during the
interval.
THINKTIME Average transaction think time (in seconds).
FIRSTRESP Average transaction time-to-first response
(in seconds).
PROMPT Average transaction response-time-to-prompt
(in seconds).
Process Wait States Metrics
STOPCPU Time process spent waiting for CPU
(percentage of total).
STOPCPU_SECONDS Time process spent waiting for CPU (in
seconds).
STOPDISC Time process spent waiting for DISC IOs
(percentage of total).
STOPDISC_SECONDS Time process spent waiting for DISC IOs (in
seconds).
STOPIMPEDE Time process spent IMPEDED
(locked)(percentage of total).
STOPIMPEDE_SECONDS Time process spent IMPEDED (locked, in
seconds).
STOPOTHERIO Time process spent waiting for OTHER IO
(percentage of total).
STOPOTHERIO_SECONDS Time process spent waiting for OTHER IO (in
seconds).
STOPSWAP Time process spent waiting for MEMORY
(percentage of total).
STOPSWAP_SECONDS Time process spent waiting for MEMORY (in
seconds).
STOPTERM Time process spent waiting for TERMINAL
input (percentage of total).
STOPTERM_SECONDS Time process spent waiting for TERMINAL
input (in seconds).
Table 5-7. Codes for Reason of Interest
The INTEREST field consists of 12 independent columns.
Each column contains either a blank or a character
representing a process INTEREST code, as shown below.
Position CharacterMeaning
1 N New (process is newly created).
2 K Killed (process is terminated).
3 C CPU percentage exceeded threshold.
4 D Disc I/Os exceeded threshold.
5 P Response-to-prompt exceeded threshold.
6 F First-response exceeded threshold.
7 T Transaction rate exceeded threshold.
8 c Wait on CPU percent exceeded threshold.
9 d Wait on disc percent exceeded threshold.
10 m Wait on memory percent exceeded threshold.
11 i Wait on impede percent exceeded threshold.
12 blank Not used.
DISC Data Type.
The following data items are available for the DISC data type.
Table 5-8. EXPORT Items for Data Type DISC SPACE
Record Identification Metrics
RECORD_TYPE ASCII field to identify this record type
"DISC".
DATE Date in MM/DD/YY format (or custom NLS
date).
TIME Time in HH:MM 24-hour format.
DAY Julian day-of-the-year (1-366).
YEAR Year (such as 1991).
DATE_SECONDS Date in UN*X format (in seconds, since
January 1, 1970).
BLANK An empty field used as a spreadsheet place
holder.
NUMBER_OF_DISCS Number of discs configured on the system.
NUMBER_OF_GROUPS Number of disc groups logged in this record.
Overall Disc Space Metrics
CAPACITY Total disc capacity on system (sectors).
TOTAL_FREE Total free disc space in sectors
(Virtual+Permanent).
LARGEST_FREE Largest contiguous free space (sectors).
Permanent Files Disc Space Metrics
FILES Total disc space used by permanent files
(sectors).
FREE_PERMANENT Total disc space available for permanent
file usage.
Table 5-8. EXPORT Items for Data Type DISC SPACE (cont.)
Transient (Virtual) Disc Space Metrics1
TRANSIENT_CAPACITY Total disc space reserved for transient or
virtual memory.
VIRTUAL_CAPACITY* Total disc space reserved for transient or
virtual memory.
PEAK_TRANSIENT Peak transient/virtual memory usage during
the last day.
PEAK_VIRTUAL* Peak transient/virtual memory usage during
the last day.
TRANSIENT Total disc space used by transient/virtual
memory.
VIRTUAL* Total disc space used by transient/virtual
memory.
FREE_TRANSIENT Total disc space available for use by
transient objects.
FREE_VIRTUAL* Total disc space available for use by
virtual memory.
Free Disc Space Fragmentation Metrics
FRAGMENT99 Total disc space in fragments <=99 sectors
in size.
FRAGMENT1K Total disc space in fragments 100-999
sectors.
FRAGMENT10K Total disc space in fragments 1,000-9,999
sectors.
FRAGMENT100K Total disc space in fragments 10,000-99,999
sectors.
FRAGMENT_OVER100K Total disc space in fragments 100,000
sectors and larger.
Permanent Disc Space User Metrics. The following fields will be repeated once for
every disc group logged.
@GROUP_NAME Name of the disc group (or account name).
@GROUP_SECTORS Sectors used for permanent files.
1 The terms TRANSIENT and VIRTUAL are used by MPE/iX and MPE V, respectively. To
allow the same report files to be used on different types of systems without
modification, the terms "TRANSIENT" and "VIRTUAL" may be used interchangeably. For
Example: "PEAK_TRANSIENT" and "PEAK_VIRTUAL" represent the same metric and may be
used on either system.
CONFIGURATION Data Type.
The following data items are available for the CONFIGURATION data type.
There is one record for each collector start-up or NEWPARM.
Table 5-9. EXPORT Items for Data Type CONFIGURATION
Record Identification Metrics
RECORD_TYPE ASCII field to identify this record type
"CONF".
DATE Date in MM/DD/YY format (or custom NLS
date).
TIME Time in HH:MM 24-hour format.
DAY Julian day-of-the-year (1-366).
YEAR Year (such as 1991).
DATE_SECONDS Date in UN*X format (in seconds, since
January 1, 1970).
BLANK An empty field used as a spreadsheet place
holder.
System Identification Metrics
SYSTEM_ID System identification string.
SYSTEM_TYPE Type of system (3000/960, etc.).
SERIAL_NO System hardware serial number (HPSUSAN,
etc.).
OP_SYS_NAME Operating system name (MPE V, MPE/iX, HP-UX
etc.).
OP_SYS_VERSION Operating system release number.
MEMORYSIZE Size of main memory (kilobytes).
USERMEMORY Amount of main memory available for
nonresident programs.
SWAPMEMORY Amount of swap memory available on disc
(HP-UX only).
NUMPROCESSORS Number of processors configured.
NUMBER_OF_DISCS Number of disc drives configured.
NUMBER_OF_DATACOMM Number of LAN interfaces (not implemented on
MPE).
Collector Identification Metrics
COLLECTOR Name and version of the Performance Data
Collection program.
LOGFILE_VERSION Version of the log file (A or B).
LOGGING_TYPES Types of performance data being logged (G,
A, P, or D).
Table 5-9. EXPORT Items for Data Type CONFIGURATION (cont.)
Interesting Process Thresholds
THRESHOLD_CPU Interesting process CPU threshold setting
(percentage).
THRESHOLD_DISC Interesting process DISC threshold setting
(IOs/second).
THRESHOLD_TRANS Interesting process TRANSACTION threshold
(number).
THRESHOLD_FIRSTRESP Interesting process FIRST RESPONSE threshold
(in seconds).
THRESHOLD_RESPONSE Interesting process RESPONSE threshold (in
seconds).
THRESHOLD_NOKILLED Interesting process Don't log KILLED
processes flag.
THRESHOLD_NONEW Interesting process Don't log NEW processes
flag.
THRESHOLD_NOSHORT Interesting process Don't log SHORT
processes (in seconds).
THRESHOLD_MINTHINK Transaction minimum think time threshold (in
seconds).
THRESHOLD_MAXTHINK Transaction maximum think time threshold (in
seconds).
WAIT_CPU Interesting process waiting for CPU
threshold (percentage).
WAIT_DISC Interesting process waiting for DISC
threshold (percentage).
WAIT_MEMORY Interesting process waiting for MEMORY
threshold (percentage).
WAIT_IMPEDE Interesting process waiting for IMPEDE
threshold (percentage).
Terminal Transaction Distribution Bounds
DISTRIBUTION_FIRST Lower limits of first-response distribution
(10 values).
DISTRIBUTION_PROMPT Lower limits of response-to-prompt
distribution (10 values).
DISTRIBUTION_THINK Lower limits of Think time Distribution (10
values).
Individual Disc Logical Device Numbers. The following field will be repeated once
for every disc on the system.
@DISC_LDEV Logical device number of disc drive.
Resulting Exported Files
By default, exported files are created with the following
characteristics. You may use file equations to override these defaults,
if you wish.
Table 5-10. Characteristics of Exported Files
Maximum Number Of Records: 50,000
Record Width: Adjusted, based on items chosen
Maximum:
MPE/iX 8000
Bytes
MPE V 4000
Bytes
Record Format: Variable
Record Type: ASCII (for ASCII and DATAFILE)
BINARY (for BINARY)
File Name: Global Data: XFERGLOB
Global Summary Data: XFERGSUM
Application Data: XFERAPPL
Application Summary: XFERASUM
Process Data: XFERPROC
Disc Space Data: XFERDISC
Configuration Data: XFERCONF
The maximum number of discs supported in the repeating fields is:
MPE/iX: 64
MPE V: 32
NOTE Currently, the MPE/iX Application Records are recorded with only
one detail disc drive. This drive contains the summary of all
other drives.
The contents of each file will be:
REPORT TITLE LINE (If REPORT Title was specified and
HEADINGS=ON.)
HEADING LINE1 (If HEADINGS=ON.)
HEADING LINE2 (If HEADINGS=ON and it is not a binary
file.)
FIRST DATA RECORD
SECOND DATA RECORD
Report title and heading lines will not be repeated in the file.
FIXED versus VARIABLE Length Record Formats.
By default, the exported files will be built with variable-length
records. If a report format has repeating fields, only the valid number
of fields will be written to the file. Using variable-length records can
save considerable disc space. If your application can not handle
variable-length records, you can override it with the following file
equation:
:FILE XFERGLOB;REC=,,F
Each record will be padded to fill the maximum record size. The records
are padded with blanks for ASCII and DATAFILES, and with binary zeros for
BINARY files.
Maximum Record Width.
Do not override the maximum record width for an exported file. If you
try to write to a record that is too small, you might abort the EXTRACT
program. If you want to truncate the records, export to the default
variable-length record file, then copy the file before truncating it.
Details on ASCII and Data File Formats.
All data in these format files should be in printable ASCII format.
ASCII and DATAFILE formats are identical except that in the latter, all
non-numeric fields are enclosed by double quotes. Even the DATAFILE
HEADER information will be quoted. ASCII file format does not use double
quotes to enclose fields, therefore these files will be more aligned when
printed. DATAFILE format is best when used for PC analysis programs.
Numerical values will be formatted based on their range of values and
internal accuracy. All fields will not be the same length so be sure to
decide how you will determine the start of each field.
The user specified SEPARATOR character (or the default blank) will be
used to separate each field from the next in ASCII and DATAFILE formats.
Blank separators can be visually more attractive if you plan to print the
report. Other separator characters may be more useful if you plan on
reading the report file with another program. A commonly acceptable
format for many PC applications is to use the comma as a separator.
SEPARATOR = ","
Be aware that some data items may contain commas that are not separators,
and this can confuse the analysis programs. The process LOGON item will
often contain a comma to delimit the job or session name and the logon
group from the user and account names. The date and time formats may
contain different special characters based on the Native Language chosen
when the EXTRACT program is run.
NOTE To enter a nonprinting special character, enter it into your report
format file following the first double quote. You might have to
turn on the DISPLAY FUNCTIONS feature to enable your terminal to
transmit the character to the file.
Hint Most spreadsheets will accept files in DATAFILE format using
SEPARATOR=" , ".
On some packages, the horizontal tab character can be used to
delimit columns. You can specify the horizontal tab character by
typing SEPARATOR= "TAB". On some terminals you might need to turn
on the DISPLAY FUNCTIONS feature to actually enter the horizontal
tab character properly.
Hint For a more attractive printout, try specifying ASCII format and the
vertical bar character (SEPARATOR="|"), then print with underlining
turned on, if you have a printer that supports this.
Details on Binary Format.
Binary format files will write numerical values as 32-bit integers. This
format can save space by reducing the overall file size, but it requires
a program that is able to read binary files. Copying a binary-format
file to a printer or a terminal is not recommended.
Non-numerical data will be written the same as it was in the ASCII format
except separator characters will not be used. To properly utilize a
binary format file, you should use the record layout report printed by
EXTRACT when you specify REPORT reportfile,SHOW. This report will give
you the starting byte of each item specified.
To maintain maximum precision and avoid nonstandard, binary
floating-point representations, all numerical values will be written as
scaled 32-bit integers. Some items may be multiplied by a constant
before they are truncated into integer format.
For example, the number of seconds of CPU time used is multiplied by 1000
before being truncated. To convert the value in the exported file back
to the number of seconds, divide it by 1000. The scale factors can be
written to the exported file for ease in conversion by specifying
HEADERS=ON. The report title, if specified, and a single header record
will be written preceding the binary data.
Binary Header Record Layout.
If the RECORD_TYPE item is selected, it will be written at the same
location as it is in the data records. The ASCII code will be the same
as in the data record, except it will be in lower case (glob, appl, proc,
disc, conf). This can be useful because it insures that you are looking
at a header record and not a report title or data record.
Consider the header record as an array of 32-bit integers. All data in
the data records are aligned to a 32-bit boundary. If the data record
contains a numerical value, then the header record will contain its scale
factor in the corresponding location. Non-numeric items will have zeros
in all corresponding locations in the header record. Negative numbers
may be used to indicate an invalid item specification. (Invalid items
should not be passed by the report file parser so negative scale factors
indicate a program bug).
Special Scale Factors.
The ASCII string, RECORD_TYPE, will appear as a very large scale factor
if you try to decode it as an integer.
The scale factor for DATE is 512. DATE format is MPE CALENDAR format in
the least significant 16 bits of the field--the rightmost 16 bits.
Scaling this as a 32-bit integer (dividing by 512) will isolate the year
as the integer part of the date, and the day of the year (divided by 512)
as the fractional part.
The scale factor for TIME is 65536. TIME is a four-byte binary field
(hour, minute, second, tenths of seconds). Dividing by 65536 will form a
number where the integer part is the (hour * 256) + MINUTE.
It may be easier to handle a DATE_SECONDS value in a binary file.