开云体育

ctrl + shift + ? for shortcuts
© 2025 开云体育
  1. H390-MVS
  2. Topics

Allocating data set - block vs track vs cylinder

 

Hi,

When using in MVS-TK5 the panel 3.2 to allocate a sequential data set

  • Record format: FB
  • Logical record length: 80
  • Physical block size: 27920
  • Volume: TSO001

Then I get to allocate in Tracks, Cylinders or Blocks.

I wanted to find out the smallest data set that I can create in an IBM 3390. When I try to allocate 1 block, DSLIST shows ALTRK=1

I thought that the minimum allocation in disks was 1 block. But, in MVS-TK5, is 1 track the smallest I can allocate?

I have tried to look for more information about it, without success. Could you clarify or point me to the a documentation that explains which component is responsible for that limit? Is it MVS, the disk controller or the disk itself?

Kind regards

Sergio Lindo Mansilla
 

Your observation is correct but the expectation is wrong. This has nothing to do with any particular DASD device type or the TK distribution of MVS.? Since OS/360 appeared or at least was announced in 1964, DASD datasets (and the VTOC, Volume Table of Contents) are allocated as an integer number of tracks. PCP, MFT, MVT, VS/1, VS/2 and so on work the same.? VTOC entries typically indicate the low and high cylinder/head combinations for each extent of the file.
Anders Edlund
 

PDS members is only as big as the data you put into it. VM was 4KB
block orientent, not sure of minimum size. DOS used VTOC without
freespace tracking and no PDSs.

On Wed, Oct 23, 2024 at 12:15?PM andersedlund via groups.io
<andersedlund@...> wrote:

Your observation is correct but the expectation is wrong. This has nothing to do with any particular DASD device type or the TK distribution of MVS. Since OS/360 appeared or at least was announced in 1964, DASD datasets (and the VTOC, Volume Table of Contents) are allocated as an integer number of tracks. PCP, MFT, MVT, VS/1, VS/2 and so on work the same. VTOC entries typically indicate the low and high cylinder/head combinations for each extent of the file.
Anders Edlund


--
Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?
 

Hi, Mike,
?
Originally, CMS on CP-67 and VM-370 supported only 800 byte physical records (block size) for CMS minidisks.
?
CP-67 and VM/370 CP supported 4K pages on disk.
?
With CP-67 and VM/370 CP, you have to manually "carve up" DASD devices into one or more "minidisks" defined to CP, in the CP DIRECTory.
?
With VM/BSEPP and SEPP and above (VM/SP, VM/ESA etc.) we got the choice of CMS minidisks using the EDF Extended Disk Format with block sizes of 512, 1024, 2048 or 4096.? ?So, depending on average file sizes, you could "tune" your CMS minidisks to eliminate "internal fragmentation."
?
DOS-360 -> DOS/VS -> DOS/VSE did support four types of "library" files (somewhat akin to PDS):? SSL (source statement library (for macros and source copy books), RL (relocatable library) for object decks for use with the linkage editor, PL (procedure library) for JCL PROCs, and CIL or CL (Core Image Library) for link-edited executable program modules.
?
DOS-360 -> DOS/VS -> early DOS/VSE did not use VTOC space management; sys.admins had to keep track of it manually.
?
With DOS/VS and above, VSAM provides its own "space management" for allocating and managing VSAM ESDS, KSDS and RRDS datasets within VSAM spaces that a sys. admin. has defined ("carved out") on DASD.
?
Mark S. Waterbury
 

On 24/10/2024 3:30 am, Sergio via groups.io wrote:
When I try to allocate 1 block, DSLIST shows ALTRK=1
Correct - for OS/360 and descendants the smallest quantum for disk space allocation is one track.

The VTOC keeps track (ha!) of which tracks are allocated to which data sets and which tracks are free (unallocated).

You can use a part of a track but you cannot allocate part of a track.

When using ISPF (the real one) and other utilities to show data set attributes, space quantities shown may not match the allocation spec used at creation time.

If you request 1 block and 10 blocks fit to the track then a 10 block allocation may be reported.

When you request 300 tracks (for example), up to 5 extents may be used to satisfy the primary space request.
If the 300 tracks were allocated in 5 extents and the first extent was 120 tracks, then IBM's ISPF option 3.2 would report the primary allocation as 120 tracks.
Note that if the primary allocation used 5 extents, then only 11 extents of secondary space are available for future use.
Multiple extents are used for the primary space allocation when there is no free extent large enough to satisfy the request in 1 extent - hence the value of defragging disks.

For MVS 3.8 the "best fit" algorithm is used, so that if you request 5 tracks and there is a 5-track free space extent, then that will be used for the request.

I forget when - probably in MVS/ESA times - the algorithm changed to "first fit".

Cheers,
Greg

1 - 5 of 5