All these "Times" are duration spent inside the Multi-user (so outside of the server).
Requests coming from Server are processed by the Multi-User.
ETIME - WTIME = RTIME (run time)
-
ETIME Elapsed time, in microseconds, from the time that the request is received by the Multi-User Facility until it is returned to the user.
-
CTIMEReflects the CPU time, in microseconds, which the Multi-User Facility spent executing the request.
CTIME adds additional operating system overhead. If you do not use this element in any of your Accounting tables, you can reduce this overhead by coding
NO for the cpu-time parameter in the
ACCTPRM MUF startup option.
-
RTIMERuntime - ETIME (elapsed time) minus WTIME (wait time) in microseconds.
If only runtime is desired, it is not necessary to define ETIME and WTIME.
-
WTIMETotal time, in microseconds, that the request had to wait because of exclusive control, I/O, and so forth.
--------------------------------------------------------------
Other relevant information:
Differences Between RTIME, CTIME, and ETIME
RTIME - Accounts for time the request is actually processed. It is inflated by page faults and it continues to accumulate even if
the Multi-User Facility has been swapped out.
CTIME - Reflects actual CPU time
ETIME - Tracks the time from initiation of a request to completion. A timestamp is taken at the moment a request starts in Multi-
User. A second timestamp is taken at the moment it finishes. The difference between these two is called ETIME.
WTIME - The amount of time a request has to wait during its execution. It contains waits that are known to Multi-User, e.g., waits for
exclusive control and I/O waits.
When WTIME is subtracted from ETIME, RTIME remains. This is not CPU time. CTIME (MVS only) comes closest to pure CPU time. It is RTIME
minus the "system" waits.
At times, CTIME can be greater than RTIME. This can occur because Multi-User overhead (index queue processing, accounting processing)
is included in CTIME.
ETIME is clocked from the time that Multi-User accepts a request until that request has completed processing in Multi-User. Because
MUF is processing many requests at once, the elapsed time of all requests in Multi-User at one time are all occurring simultaneously.
While a task is waiting on I/O or on a higher priority task, instructions are not executing, but time is passing.
CTIME represents our best efforts at charging for instruction execution to the request being processed. Multi-User may have many requests work-
ing at once, but at one time, only one is executing instructions and therefore, that one is charged. This will vary depending on the
workload being done.
CTIME can be more than ETIME when Multi-User is running on a multi-processor. Because we have more TCBs in a MUF running and those TCBs
can be run on separate processors, part of this 'MUF overhead', running under other TCB's than the TCB belonging to the specific task, is
included in the CTIME for that task.
The CTIME spent ending one request and waiting for the next is charged to the caller on the next call when possible.
-------------------------------------------------------------------
There is a Video Webcast related to Accounting at the following address:
https://support.ca.com/us/product-content/recommended-reading/technical-document-index/ca-datacom-video-and-recorded-webcast-index.html?intcmp=searchresultclick&resultnum=7?intcmp=searchresultclick&resultnum=7