When creating an indexed sequential data set, you can improve performance by using the full-track-index write option. You do this by specifying OPTCD=U in the DCB. OPTCD=U causes the operating system to accumulate track index entries in virtual storage. The full-track-index write option can be used only for fixed-length records.

If you do not specify full-track-index write, the operating system writes each normal overflow pair of entries for the track index after the associated prime data track has been written. If you do specify full-track-index write, the operating system accumulates track index entries in virtual storage until either (a) there are enough entries to fill a track or (b) end-of-data or end-of-cylinder is reached. Then the operating system writes these entries as a group, writing one group for each track of track index. The OPTCD=U option requires allocation of more storage space (the space in which the track index entries are gathered), but the number of I/O operations required to write the index can be significantly decreased.

When you specify the full-track-index write option, the track index entries are written as fixed-length unblocked records. If the area of virtual storage available is not large enough the entries are written as they are created, that is, in normal overflow pairs.

Example: The example in Figure 1 shows the creation of an indexed sequential data set from an input tape containing 60-character records.

//INDEXDD DD     DSNAME=SLATE.DICT(PRIME),DCB=(BLKSIZE=240,CYLOFL=1,    C
//               DSORG=IS,OPTCD=MYLR,RECFM=FB,LRECL=60,NTM=6,RKP=19,    C
//               KEYLEN=10),UNIT=3380,SPACE=(CYL,25,,CONTIG),---
//INPUTDD DD     ---
          ...
ISLOAD    START  0
          ...
          DCBD   DSORG=IS
ISLOAD    CSECT
          OPEN   (IPDATA,,ISDATA,(OUTPUT))
NEXTREC   GET    IPDATA                 Locate mode
          LR     0,1                    Address of record in register 1
          PUT    ISDATA,(0)             Move mode
          B      NEXTREC
          ...
CHECKERR  L      3,=A(ISDATA)           Initialize base for errors
          USING  IHADCB,3
          TM     DCBEXCD1,X'04'
          BO     OPERR                  Uncorrectable error
          TM     DCBEXCD1,X'20'
          BO     NOSPACE                Space not found
          TM     DCBEXCD2,X'80'
          BO     SEQCHK                 Record out of sequence

IPDATA    DCB    ---
ISDATA    DCB    DDNAME=INDEXDD,DSORG=IS,MACRF=(PM),SYNAD=CHECKERR
          ...

The key by which the data set is organized is in positions 20 through 29. The output records will be an exact image of the input, except that the records will be blocked. One track per cylinder is to be reserved for cylinder overflow. Master indexes are to be built when the cylinder index exceeds 6 tracks. Reorganization information about the status of the cylinder overflow areas is to be maintained by the system. The delete option will be used during any future updating.