IBM Mainframe Forum

Hi,

Please assit me by providing clarification as i am pretty confused with dataset extensions.

PDS :

1. If i mention PRIQTY 10 and SECQTY 10, Directory Block 2, Allocation Form is CYL and RECL=80 RECF=FB --> So how the size will get calculated, how many extends for the secondery, how the extends will be allocated.

2. If i mention PRIQTY 10 and SECQTY 10, Directory Block 2, Allocation Form is CYL and RECL=80 RECF=VB --> So how the size will get calculated, how many extends for the secondery, how the extends will be allocated.

Finally How many members the PDS can have ?

LIBRARY :

1. If i mention PRIQTY 10 and SECQTY 10, Directory Block 2, Allocation Form is CYL and RECL=80 RECF=FB --> So how the size will get calculated, how many extends for the secondery, how the extends will be allocated.

2. If i mention PRIQTY 10 and SECQTY 10, Directory Block 2, Allocation Form is CYL and RECL=80 RECF=VB --> So how the size will get calculated, how many extends for the secondery, how the extends will be allocated.

Finally How many members the PDS can have ?


Please clarify.

Regards,
Nantha.Y

In general, the DCB attributes are independent of the space allocation.

1. In JCL you are specifying SPACE=(CYL,(10,10,2)). In theory, the initial allocation will be 10 cylinders, but the allocation may be broken into up to 5 extents. You have no control over this. The entire PDS directory must fit into the first extent, though that is rarely an issue. All secondary allocations will be 10 cylinders, up to a total of 16 extents.
2. Max members. There are two limits, and they are independent of each other.
   • Directory capacity. A directory entry in a PDS potentially varies in length from 12 to 74 bytes. Each directory block contains 254 bytes that can be used for directory entries. There is an unbending requirement for 1 minimum length directory entry in every directory.
     • A directory entry for a text member without ISPF member statistics requires 12 bytes. In your data set this means 41 members. It is possible to create a longer directory entry, but most users don't have a clue about how to do this. In 40+ years, I've never done this.
     • A directory entry for a text member that that has ISPF member statistics contains 42 bytes. In your data set this means 11 members.
     • Directory entries for a load module truly are variable, but most require 36 bytes, and most users do not know how to create longer entries. In your allocation this means 13 members.
   • Max data size. This depends on the data; this is something you must determine yourself. The largest PDS can contain up to 65535 tracks, though the largest PDS with your allocation parameters will be much smaller.

The MVS JCL Reference manual for your z/OS release is the best reference for most of this, though it does not have any information about directory entry size that is in this post. See the "System Assignment of Space" section, though that can be hard to find since it is not in the main table of contents; it is in the "SPACE Parameter" chapter which is in the main table of contents.

Hi Steve,

Thanks for the information.

One quick question on initial space allocation, Let say, i am going to create dataset with size of primary allocation is 300 CYL with LRECL = 80 but assume that i may not be use this dataset at all or dataset going to be empty for long time.

Size information is : LRECL = 80 , SPACE = (CYL,(300,2))

In this case, how IBM Mainframe will allocate the initial space, will it allocate entire 300 CYL for this purpose or just some KBs until the dataset going to get some values ( at least one character).

Could you please advise on this scenario ?

Regards,
Nantha.Y

When you allocate a data set on disk, the primary space is allocated -- whether or not there is any data in the data set. Depending upon your site's storage management policies and procedures, such a data set may be left as is, or it may be reduced to some minimum size after some minimum period of time. You would need to talk to your site's storage management group to know for sure, and to find out what the time frame is as well as how small the data set would become.

Hello,

Do keep in mind that dataset allocation is quite different on the mainframe as compared to Win-based and *nix platforms.