o It is usually much faster to use native operating
system commands for copying & deleting
datasets, catalogs, etc. You can use the new
version 8 command to execute o/s commands
asynchronously in order to save even more time.
o It is advisable to clean up work space during a
SAS session since SAS only does so at the end.
Particularly large work datasets should be
deleted by you when they are no longer needed,
otherwise they may take valuable space
o You should clean upmemory every so often, by
using the “CDE P” command to purge unused
modules from memory.
o You can split data into smaller bits for sorting.
This is especially useful if they can fit in
memory. In one of my examples I saved 36% of
the elapsed time.
o You can use compressed single keys made up of
several variables for sorting and indexing. This
usually takes less time and results in your
dataset being in the same order.
o You should try to reduce I/O contention by
putting work, swap and data on different
physical disks where possible.
No comments:
Post a Comment