Optimizing Easytrieve VIRTUAL File Translations in IMU: Minimizing Work Spool DASD

During an enterprise migration from Broadcom Easytrieve Classic to IBM Migration Utility (IMU / EASYTRAN), converted batch jobs experience severe DASD consumption:

IEC030I B37-04,IFG0554A,JOBE012,STEP01,FSRTWORK,1402,PROD01,SYS26281.T091214.RA000.JOBE012.R0100001
FSORT001E WORK DATA SET FULL. UNABLE TO COMPLETE VIRTUAL SORT.

Jobs that previously executed in 4 minutes with native Easytrieve run for 35 minutes under IMU, allocating up to 80GB of temporary scratch space across JES spool and DFSMS volumes.

Easytrieve programs frequently declare:

FILE VIRTFILE VIRTUAL

Native Easytrieve manages small virtual files in below-the-bar virtual storage. When IMU translates VIRTUAL files into generated COBOL, its default configuration directs records to intermediate temporary disk files (FSRTWORK / SYSUT1) and triggers physical DFSORT work datasets unless explicit in-memory options are enabled in EASYTRAN options.

# 1. Adjust EASYTRAN Compiler Options

In the IMU translation step, pass the MEMORY and VIRTUAL=MEMORY parameters:

//TRANSLAT EXEC PGM=FSYACC00,
// PARM=('EASYTRAN',
//       'VIRTUAL=MEMORY',
//       'FSORT=INTERNAL',
//       'NOWORKFILE')

• VIRTUAL=MEMORY: Instructs IMU to generate in-memory COBOL working storage tables with OCCURS DEPENDING ON rather than temporary sequential datasets.
• FSORT=INTERNAL: Leverages DFSORT hiperspaces and memory buffers rather than physical DASD work units.

# 2. Configure 64-Bit Memory Allocation in Generated COBOL

Ensure the target Enterprise COBOL compiler step utilizes 64-bit or Extended Addressing (AMODE 31/64):

//COMPILE EXEC PGM=IGYCRCTL,
// PARM=('ARCH(13)','ARITH(EXTEND)','OPT(2)','LP(64)')

# 3. Generated COBOL Architecture Comparison

Instead of physical I/O writes:

* DEFAULT UNOPTIMIZED TRANSLATION:
       WRITE VIRTUAL-REC TO FSRTWORK-FILE.

      * OPTIMIZED IN-MEMORY TRANSLATION:
       ADD 1 TO WS-VIRTUAL-INDEX
       MOVE INPUT-DATA TO WS-VIRTUAL-TABLE(WS-VIRTUAL-INDEX).

This reduces execution time by up to 88% and eliminates temporary DASD allocation entirely!