"The?disks are formatted?into 800-byte?physical records,?called?blocks. Logical records,?which may be?fixed-length or variable-length, are?imposed on?constant physical blocks. Space required for?files is?automatically allocated?by CMS. As a?file?grows,?its?space?is?expanded,?and?it?is?contracted as?its?space requirements are reduced."
Also, according to the chart on page 27, only CKD disks are supported for VM/370.
On Sat, Feb 1, 2020 at 3:12 PM George Shedlock <gshedlock352@...> wrote:
Steven,
If you are referring to VM minidisks
(such as those available to a cms user), I believe that the device
type is always the same as what the VM operating system thinks
they are. So if VM thinks the "real" device is ckd, then the cms
minidisk is also ckd.
George Shedlock
On 2/1/2020 4:02 PM, Dave Wade wrote:
Steven,
I haven¡¯t looked at the code, but VM/370R6 only
uses 800byte blocks so CKD for minidisks. I don¡¯t know how
this is mapped to clusters on windows or linux, but its
possible it overlaps two physical sectors¡
Later versions of CMS use 512/1024/2048/4096 (I
think) block sizes which could map to physical sectors, but
I bet there is some overhead. On an SSD its probably very
low.
Related
to this discussion of performance is the virtual
device associated with a minidisc always the same type
of device (CKD or FBA) as the underlying device or can
it be different.? If it can be different is there a
performance penalty for conversion?