Direct access storage devices

Each record that is written on direct access storage devices requires some device overhead.

Use the TRKCALC macro to calculate the exact number of bytes required for each data block including the space that is required for device overhead. For more information on how to use the TRKCALC macro, see z/OS DFSMSdfp Advanced Services.

If the TRKCALC macro cannot be used and space calculations must be performed manually, refer to the appropriate Direct Access Storage Reference Summary.

The following tables help you estimate your space needs.

Table 1 lists the physical characteristics of DASDs. Today, disk storage subsystems emulate the track capacity of an IBM 3380 or 3390 device while providing much larger capacity than the original 3380 and 3390 devices. For example, the IBM System Storage DS8000 series emulates emulates the IBM® 3390. On an emulated disk or on a VM minidisk, the number of cylinders per volume is a configuration option. It might be less than or greater than the stated number. If so, the number of bytes per device differ accordingly. The IBM ESS Model 2105 supports up to 65520 cylinders and the IBM DS8000® supports up to 1,182,006 cylinders.

In a compressed format data set, PDSE z/OS UNIX file or dummy data set, the most efficient block size is unrelated to the track length. The most efficient block size would be up to 32760.

The maximum data length for a track multiplied by the number of tracks per cylinder produces the number of bytes available per cylinder for a device.

Similarly, the number of bytes per cylinder multiplied by the number of cylinders per volume produces the total number of bytes available for a device.

Table 1. DASD Physical Characteristics
Type	Most Efficient Block ¹	Maximum Data Length/Track²	Trk/Cyl	Bytes/Cyl Avail for User Records³	Cyl/Vol
3390 emulation using IBM DS8000 device	27,998	56,664	15	849,960	1 – 1,182,006
3380 emulation using IBM DS8000 device	23,476	47,476	15	712,140	1 – 1,182,006
3390 Model 1	27,998	56,664	15	849,960	1,113
3390 Model 3	27,998	56,664	15	849,960	3,339
3390 Model 9	27,998	56,664	15	849,960	10,017
9345 Model 1	22,928	46,456	15	696,840	1,440
9345 Model 2	22,928	46,456	15	696,840	2,156

Note:

This assumes standard access methods writing two blocks per track. With an extended format data set that is not compressed format, the most efficient block size is 32 less.
This is the theoretical maximum length of one block. The access methods do not support such long blocks.
This is a theoretical one very long block per track. If you use the standard access methods, multiply 30 times the most efficient block size.