This document covers the following topics:
A GSAM database uses the Generalized Sequential Access Method
(GSAM) under DL/I. It is recognized by the
'ACCESS=GSAM' keyword for the DBD statement in the DBD definition.
A GSAM DBD does not specify any segment or field definitions. The
corresponding PCB statement in the PSB definition uses
'TYPE=GSAM'.
For more information regarding the definition of a GSAM DBD
refer to the IMS Utilities Reference Manual of IBM.
GSAM databases are supported by ADL with some restrictions
regarding the DBD definition, the JCL and the OPEN command.
In particular, the following restrictions are valid for a GSAM
DBD:
The records must have fixed length, i.e. the RECFM
keyword of the DATASET statement in the DBD definition must have a
value of 'F' or 'FB'.
The record length (RECORD keyword) must be at least 12
bytes.
If these conditions are met, the GSAM DBD can be converted
and the data can be loaded into Adabas, as described in the next section.
Like for any other converted database, the data set corresponding to a
converted GSAM database cannot longer be referenced in the JCL, because the
data is in Adabas. Consequently, the record format
(RECFM) parameter cannot be overwritten in the JCL. With
DL/I the disposition (DISP) of the output dataset defines, whether
a new dataset is created or whether the data is added to the end of the
dataset. With ADL the data is always added to the end. If you want to insert
data from the beginning of a data set, which does already contain data, you
must refresh the Adabas file before your program starts. This can be performed
by either using the REFRESH function of the Adabas
ADADBS utility or manually with the Adabas Online System.
The DL/I OPEN and CLSE calls are ignored by
ADL. Under DL/I you can add an option to the OPEN call, which
specifies the kind of control character in the first byte of each record in the
output data set. With ADL there is always no control character in the first
byte of each output record.
With DL/I an application can use the CLSE and
OPEN commands to read a GSAM database after it was
loaded, without terminating in between. This works in the same way after the
migration, because ADL does not need these calls for a re-positioning.
Run the GSAM DBD through the step 1 (assembly), step 2 (CBC
utility) and step 3 (create input decks) of the ADL control block conversion
(CBC). For a more detailed description of the ADL CBC, see the section
ADL Conversion Utilities for DBDs and PSBs in this
documentation. Step 4 (assemble and link-edit Adabas user 6 extension) is not
required for a GSAM database.
ADL adds a segment 'GSAMROOT' and a sequence field
'GSAMFLD' to the DBD definition. The length of the segment is the
same as specified with the 'RECORD' keyword in the
DATASET statement. The field always has a length of 12 bytes.
Additionally, ADL adds a sensitive segment 'GSAMROOT' to the
GSAM PCB.
The ADACMP cards which are generated by step 3 of the ADL
conversion utility is used as input for the ADACMP COMPRESS
utility. Note, that the USERISN=YES option must not be used for
any Adabas utility run against a file of a converted GSAM DBD.
There is no need to run the ADL unload utility when unloading the DL/I
GSAM database. Instead, the sequential file can directly be used
as input for the Adabas compression utility, by specifying the
GSAM dataset as DDEBAND for ADACMP. You
must not use the ADL user exit 06 for the compression.
Use the following parameter settings for ADALOD when
loading the data into Adabas:
DATAPFAC=1 DSREUSE=NO ISNREUSE=NO USERISN=NO
The GN, ISRT and GU calls are
supported in the same way as under DL/I. GN and ISRT
can be specified with or without a record search argument (RSA).
With ADL the first 4 bytes of the RSA contain the Adabas ISN of
the record and the last 4 bytes contain hexadecimal zeros. After a successful
call, the RSA is also placed into the Key Feedback Area of the
corresponding PCB. Further to these database calls, the system service calls
CHKP and XRST are supported.
With the LOAD parameter of ADL, you can specify
where the data of an ISRT call against a PCB with
PROCOPT=L or LS is to be written. See the section
ADL Parameter
Module in the ADL Installations
documentation for more information about the LOAD
parameter. When you specify 'LOAD=UTILITY' the data is written to
a variable blocked sequential file. For a GSAM database, this file
can be loaded back into Adabas in the same way as described above. Especially
you must set USERISN=NO, and you must not use the ADL user exit
6.
When 'LOAD=DIRECT' is specified as ADL parameter,
you can insert and read data simultaneously in one application by using
different PCBs in the PSB. In particular you do not need to issue a
CLSE and an OPEN call, when you want to retrieve data
after inserting it. You can also mix the ISRT and GET
calls as you like, and the position is kept for each PCB.
Further to inserting and retrieving data, you can also replace and
delete records, by using the REPL (replace) and DLET
(delete) calls. The corresponding PCB must specify PROCOPT=R or
=D, respectively. The data storage of a deleted record is not
released, so that a new inserted record is put to the end of the database, as
expected for GSAM databases.
Note:
An Adabas backout after a delete request acts like a re-insert, i. e.
the record is inserted to the end of the database. After such a delete/backout
the data will be therefore in another sequence than before.
After the data conversion you can manipulate it with Natural. This includes update and delete requests. Because there is no ADL internal pointer field stored with the data, there is no need to use the ADL Consistency.
ADL translates a get request against a GSAM DBD to an
Adabas L2 (read physical) call. For a high performance it is
recommended to use the Adabas Multifetch feature for a GN sequence
against a converted GSAM database. Refer to Using the Adabas
Multifetch Feature, section
Performance
Considerations in the ADL Interfaces
documentation for more information about the usage of the Adabas Multifetch
feature against migrated data. In particular, you should make sure, that the
correct Adabas command ID is prefetched.
