Allocating Space

An indexed sequential data set has three areas: prime, index, and overflow. Space for these areas can be subdivided and allocated as follows:

Prime area—If you request only a prime area, the system automatically uses a portion of that space for indexes, taking one cylinder at a time as needed. Any unused space in the last cylinder used for index will be allocated as an independent overflow area. More than one volume can be used in most cases, but all volumes must be for devices of the same device type.
Index area—You can request that a separate area be allocated to contain your cylinder and master indexes. The index area must be contained within one volume, but this volume can be on a device of a different type than the one that contains the prime area volume. If a separate index area is requested, you cannot catalog the data set with a DD statement.
If the total space occupied by the prime area and index area does not exceed one volume, you can request that the separate index area be imbedded in the prime area (to reduce access arm movement) by indicating an index size in the SPACE parameter of the DD statement defining the prime area.

If you request space for prime and index areas only, the system automatically uses any space remaining on the last cylinder used for master and cylinder indexes for overflow, provided the index area is on a device of the same type as the prime area.
Overflow area—Although you can request an independent overflow area, it must be contained within one volume and must be of the same device type as the prime area. If no specific request for index area is made, then it will be allocated from the specified independent overflow area.
To request that a designated number of tracks on each cylinder be used for cylinder overflow records, you must use the CYLOFL parameter of the DCB macro. The number of tracks that you can use on each cylinder equals the total number of tracks on the cylinder minus the number of tracks needed for track index and for prime data. That is:
```
   Overflow tracks = total tracks
                   - (track index tracks + prime data tracks)
```
When you allocate a 1-cylinder data set, ISAM reserves 1 track on the cylinder for the end-of-file mark. You cannot request an independent index for an indexed sequential data set that has only 1 cylinder of prime data.

When you request space for an indexed sequential data set, the DD statement must follow several rules, shown as follows and summarized in Table 1.
Space can be requested only in cylinders, SPACE=(CYL,(...)), or absolute tracks, SPACE=(ABSTR,(...)). If the absolute track technique is used, the designated tracks must make up a whole number of cylinders.
Data set organization (DSORG) must be specified as indexed sequential (IS or ISU) in both the DCB macro and the DCB parameter of the DD statement.
All required volumes must be mounted when the data set is opened; that is, volume mounting cannot be deferred.
If your prime area extends beyond one volume, you must specify the number of units and volumes to be spanned; for example, UNIT=(3380,3),VOLUME=(,,,3).
You can catalog the data set using the DD statement parameter DISP=(,CATLG) only if the entire data set is defined by one DD statement; that is, if you did not request a separate index or independent overflow area.

As your data set is allocated, the operating system builds the track indexes in the prime data area. Unless you request a separate index area or an imbedded index area, the cylinder and master indexes are built in the independent overflow area. If you did not request an independent overflow area, the cylinder and master indexes are built in the prime area.

If an error is found during creation of a multivolume data set, the IEHPROGM utility program should be used to scratch the DSCBs on the volumes where the data set was successfully allocated. You can use the IEHLIST utility program to determine whether part of the data set has been allocated. The IEHLIST utility program also determines whether space is available or whether identically named data sets exist before space allocation is attempted for indexed sequential data sets. These utility programs are described in z/OS DFSMSdfp Utilities. Table 1 lists the criteria for requesting indexed sequential data sets.

Table 1. Requests for indexed sequential data sets
Index Size Coded?	Restrictions on Unit Types and Number of Units	Resulting Arrangement of Areas
n/a	None	Separate index, prime, and overflow areas. See Specifying an Independent Overflow Area.
n/a	None	Separate index and prime areas. Any partially used index cylinder is used for independent overflow if the index and prime areas are on the same type of device. See Specifying a Separate Index Area.
No	None	Prime area and overflow area with an index at its end. See Specifying a Prime Area and Overflow Area.
Yes	Prime area cannot have more than one unit.	Prime area, imbedded index, and overflow area. See Specifying a Prime Area and Overflow Area.
No	None	Prime area with index at its end. Any partially used index cylinder is used for independent overflow. See Prime Data Area.
Yes	Prime area cannot have more than one unit.	Prime area with imbedded index area; independent overflow in remainder of partially used index cylinder. See Prime Data Area.