2. USAGE GUIDELINES › 2.1 Work File Compression

2.1 Work File Compression

The SAS system gives you the option to create variable-length
or fixed-length observations in a SAS data set.
Variable-length observations differ from fixed-length
observations in that the former are usually smaller because
the blank spaces used to pad fixed-length observations are
removed.
 
You instruct SAS to create variable length observations by
specifying the COMPRESS= option.  SAS data set compression
can be implemented for individual data sets or across the
entire SAS system by specifying COMPRESS= on either a DATA
statement (for the individual data set named on the DATA
statement) or an OPTIONS statement (for the entire system).
 
For more information on the COMPRESS= option, see the SAS
Institute documentation.


In deciding whether or not to implement compression for a SAS
data set, CA's research indicates the following:

 o Compressed data sets generally require fewer IOs than
   uncompressed data sets.

 o Observations in a compressed data set cannot be accessed
   by observation number.

 o Compressed data sets use more TCB CPU time than
   uncompressed data sets.  (TCB time is the amount of time
   spent executing application code.)

The IMS Analyzer supports compression for files in the
CA MICS database as well as the work files used during daily
operational processing.  If you choose to implement data
compression for the IMS Analyzer's work files, CPU and
elapsed times may be prolonged in the DAY060 step of the
DAILY job.

Before implementing work file compression, try using the
following tailoring methods to decrease the size of you IMS
Analyzer work files:

 - deactivate unused data elements following the instructions
   in Section 10.1 of this guide.

 - configure your unit databases so that products with large
   databases are in different units.  Remember that
   compression takes place at the unit level, so all products
   in a unit are compressed when compression is activated.