TPDC Manual (G06.24+)

Using the SPAM Performance Data Charts and
Reports
Tandem Performance Data Collector (TPDC) Manual—526357-001
5-77
Server Summary (from $RECEIVE File) (SERVICE)
Parameters used are:
•
LID—specifies the Measure Load ID
•
SECTION—specifies the report section number
The report is sorted by process.
Server Summary (from $RECEIVE File) Report for System \P20ET (C30)
------------------------------------------
<-ESTIMATED--> <----AVERAGES PER TRANSACTION-->
Srvice Utili- Fil <------Milliseconds----->
<-[HH:MM]-> Elaps Trans/ Av RCV Time zation Opn Logicl CPU CPU Memory I/O
Procss PID Pri Prgrm File Name From To Scnds Second Q msec (msec) (pct) Cnt I/O's Busy
Wait Wait Wait
====== --- --- --------------- ===== ----- ----- ------ ------ ------ ------- --- ------ ----- ---
--- ------ ----
(1) (2) (3) (4) (5) (6) (7) * (8) (9) (10)?/X (11) (12) (13) (14)
(15) (16) (17)
THIS REPORT SUMMARIZES SERVER PROCESS (10) The estimated average service time of the
INFORMATION. IN PARTICULAR, IT SHOWS THE process; that is, the estimated average time,
AVERAGE TIME THAT REQUESTS FOR SERVICE SPENT in milliseconds, between the time the process
WAITING IN PROCESSES’ MESSAGE INPUT QUEUES, begins work on a new transaction transaction
AND THE ESTIMATED SERVICE TIMES AND and the time it completes work on that
UTILIZATION LEVELS OF SERVER PROCESSES. transaction. This estimate INCLUDES the
possible effects of CPU queueing, memory page
faults, and the queueing and service times of
other processes that service this process’
SERVER PROCESSES ARE ASSUMED TO GET TRANSACTION logical I/O requests.
REQUESTS FROM THEIR EXTERNAL MESSAGE QUEUES
(THAT IS, FROM THEIR '$RECEIVE' FILES). N.B. CARE SHOULD BE TAKEN WHEN EVALUATING THIS
---- ESTIMATE, since it assumes that the process
handles (and completes) incoming requests one
at a time, using 'waited' I/O; the estimate
1) The Guardian process name of the server will be questionable ('?' follows) or invalid
process. ('X' follows) if:
a) The above is not true, or
2) The process identification (<cpu>,<pin>). b) The report could not match the process to
its open of its $RECEIVE file (FILE
3) The execution priority of the process. $RECEIVE not MEASURE'd, or [PARAM] ALLOW
value of 5.00 seconds is too small to
4) The server program object file name. This allow process-to-file matching in this
report shows information for one or more MEASURE data).
running copies of the program ('processes'); Some knowledge of the application is
see note (8). desirable to verify that the estimate provided
here is plausible.
5) The time at which measurement of the
process began. (11) The estimated utilization of the process,
expressed as a percentage of the report
6) The time at which measurement of the 'Window', NOT of the life of the process
process ended. itself; based on estimated service time
[see (10)].
7) (6) minus (5), to the nearest second. An
asterisk (*) following the value FLAGs that (12) The total number of logical file opens
this time is less than 90% of the report matched to the process. A large value may
'Window'. indicate the (normally) poor practice of
repeatedly opening, accessing and then
8) The number of transactions per second closing files.
handled by the process; this report EXCLUDEs
processes that handled less than 0.10 COLUMNS (13) - (17) SHOW AVERAGE VALUES PER
transactions per second. TRANSACTION:
9) The average time, in milliseconds, that (13) Logical I/O requests to MEASURE'd files.
requests for services waited in the process’
message queue. (14) Milliseconds spent actively using the CPU.
If this is a PATHWAY server process, there (15) Milliseconds spent waiting to use the CPU.
may be additional queueing of requests for
this process WITHIN one or more TCPs; a TCP (16) Milliseconds spent waiting for Memory Manager
will typically not SEND a request to a service.
server process that it knows is busy
servicing a request on its behalf. (17) Milliseconds spent waiting for completion of
logical I/O requests to MEASUREd files. This
excludes any time spent waiting for 'any file'
nowait I/O's to complete.