Considerations for CA Datacom/DB Native Access › Sequential Processing

Sequential Processing

CA Ideal invokes the CA Datacom/DB GETIT and GSETL commands for optimum batch performance (GETIT and GSETL are not invoked for on-line processing) in instances where this optional processing method is specified. Better performance is attributed to optimized record retrieval due to the following conditions:

• Multi-blocked read ahead if traversal is by key.
• Blocked record transfer between Multi-User Facility (MUF) and the CA Ideal batch application.

The following conditions must exist before GETIT and GSETL sequential processing is invoked:

• Traversal must be by a key that is no more than 90 bytes in length.
• All records in the key range must satisfy the WHERE criteria.
• Internal data retrieval order must match the default order or the order explicitly specified in an ORDERED BY clause. You cannot specify UNIQUE in an ORDERED BY clause. You cannot specify the DESCENDING keyword.
• Record count in a FOR FIRST n statement must be greater than 10.
• WHERE clause must not contain the clauses NOT EQUAL, CONTAINS, or OR.
• Parameters in the active User Requirements Table (URT) required for GETIT are listed on the following page.

CA Ideal can use the multi-block read sequential processing capabilities of CA Datacom/DB if the data is organized where the nature key can retrieve records (the key that determines the physical order of the data when the data was loaded). This type of processing uses the CA Datacom/DB GSETL and GETIT commands. Processing is restricted to a batch environment and limited to the prime data area (the area initially written during the LOAD function). The GETIT command retrieves records sequentially, starting from the record marked by the GSETL command. Multi-block reads greatly reduce the number of EXCPs required to read a table. The following also impact sequential processing:

• GETIT and GSETL acquire locks on records by blocks. To release those locks, the entire block must be processed. We recommend that you use this only with a FOR EACH/ALL construct and that you do not include any QUIT statement in the FOR construct. This is a problem only if the URT entry for the file specifies UPDATE=YES.
• If an application has substantial add and delete activity, frequently reorganizing the table keeps the data in native key sequence and maintains efficiency.
• Examine the space management option specified for the data area against the application processing requirements for the site. Option 0 and Option 2 are the most efficient for GETIT processing. (The CA Datacom/DB Database and System Administrator Guide fully describes the data area space management options.)

To implement sequential processing of a CA Datacom/DB database in CA Ideal running under batch, an appropriate User Requirements Table (URT) is required.

The following DBURTBL macro parameters, required for sequential access, are described in the CA Datacom/DB Database and System Administrator Guide.

• ELMCHG=
• GBMAXR=
• GETBLK=
• SEQBUFS= Number of data buffers the program can use for block read‑ahead sequential processing. Specify an even number of buffers.
• UPDATE=YES Code this only if updates are required, since this causes CA Datacom/DB to acquire locks on the records.

The following example illustrates the parameter values for a URT for a CA Ideal batch job doing sequential processing:

TITLE 'URT FOR EMPLOYEE DVW FOR GSETL/GETIT'

DBURSTR
      CSECT=GETITURT,
      TXNUNDO=NO
DBURTBL
      DBID=1,
      SYNONYM=YES,
      ELMCHG=NO,
      SEQBUFS=8,
      GETBLK=12288,
      GBMAXR=160,
      TBLNAM=PMF,
      UPDATE=YES
DBUREND
      USRINFO=IDEAL-BATCH-URT
END

Before issuing the RUN command in CA Ideal batch, the following command is required:

SET RUN URT GETITURT

GETITURT is the value of the CSECT specified in the CSECT= parameter above. It must match the load module name of the URT. See the Command Reference Guide for more information about the SET RUN URT command.

To verify that GSETL/GETIT commands are used, trace the batch run using the CA Ideal Dial Trace with code "V" and limit the run of a test program with SET RUN LOOPLIMIT 5 (enough to see several GETIT commands). The first request to the database is a SELFR command. CA Datacom/DB sets a flag in the request area that tells CA Ideal whether GSETL/GETIT commands can be used. If this occurs, dial trace output shows "GETIT MODE ACCEPTED." For more information about using the trace facility, see the Problem Determination Guide.