The %ARMJOIN macro reads
the six SAS data sets that are created by the %ARMPROC macro. It merges
the information from those data sets to create data sets and SAS views
for easier reporting of ARM data.
Note: The %ARMJOIN macro does not
work from SCL. It must be run in a DATA step.
The
input is the SAS data sets from the %ARMPROC macro. You must run the %ARMPROC
macro before running the %ARMJOIN macro.
The
output is a single SAS library that contains the following:
-
information about applications
(APP)
-
a DATA step view that contains
information about all start handles, including parent correlator class
and parent start handles (TXNVIEW)
-
a SAS view that contains information
about all update transactions (UPDTVIEW)
-
one transaction data set for
each application
-
one update data set for each application
The application data
set is named APP and contains one observation for every application
that is found in the input data. Each observation contains information
such as application name, user ID, transaction counts, average application
response time, and so on. In addition, each observation contains a
numeric variable APPNO that is the identifier of the related transaction
or update data set that contains more detailed transaction information.
The transaction data
sets are named TXN1, TXN2, TXN3, and so on. Each data set corresponds
to a single application, and each observation represents a single
ARM transaction containing start and stop times, elapsed times, and
CPU time.
The TXNVIEW view joins
all transaction data sets into a single data set. Start handle elapsed
time and CPU time are calculated from the start and stop transactions.
If the start handle has a parent start handle, then the class ID and
start handle of the parent are included using the variables PARCLS=
and PARHDL=. If no parent is specified, these variables contain missing
values.
The update data sets
are named UPDT1, UPDT2, UPDT3, and so on. Each data set corresponds
to a single application, and contains multiple observations for each
ARM transaction. Each observation contains the ARM call datetime,
an ARM call sequence ID, and, if applicable, elapsed time, CPU time,
and update data.
The UPDTVIEW view joins
all update data sets into a single data set.
The transaction data
sets are easier to use for analyzing individual ARM transactions because
all information about a transaction is represented in one observation.
However, the transaction data sets do not contain any information
from %ARMUPDT macros.
The update data sets
are similar to the transaction data sets. However, information about
a single transaction is spread over several observations. Update data
sets contain logged data buffer information from all %ARMUPDT macros.