Compressed SPD Engine
data sets occupy blocks of space on the disk. The size of a block
is derived from the IOBLOCKSIZE= table option that is specified when
the data set is created. When the data set is updated, a new block
fragment might need to be created to hold the update. More updates
might then create new fragments, which, in turn, increases the number
of
I/O operations needed to read a data set.
By increasing the block
padding in certain situations where many updates to the data set are
expected, fragmentation can be kept to a minimum. However, adding
padding can waste space if you do not update the data set.
You must weigh the cost
of padding all compression blocks against the cost of possible fragmentation
of some compression blocks.
Specifying the PADCOMPRESS=
table option when you create or update a data set adds space to all
of the blocks as they are written back to the disk. The PADCOMPRESS=
setting is not retained in the data set's metadata.