• Print |
• E-mail  |
• Bookmark

In the document "What's New in SAS Software for Release 8.2" under the
category of System Options, it states the following:

  When a request is made to compress a SAS data set with the COMPRESS=
  system OPTION, SAS now determines whether the resulting compressed
  file will be larger than an uncompressed file.  If so, SAS creates an
  uncompressed data set and issues a warning message stating that the
  file has not been compressed.

This statement is misleading.  The overhead associated with a compressed
data set is 12 bytes per observation.  If, due to the number and type of
the variables defined in the data set, it is not possible for the
compressed data set to be at least 12 bytes per observation smaller than
an uncompressed version, compression is disabled and a message is
written to the SAS Log.

Here is an example where SAS determines that it is not possible for
the compressed data set to be smaller.  Compression is disabled:

   data one(compress=yes);
     length x y $2;
     input x y;
   cards;
   ab cd
   ;
   run;

The following message is written to the SAS Log:

   WARNING:  Compression was disabled for data set WORK.ONE because
             compression overhead would increase the size of the data
             set.

Because both of the SAS-supplied compression algorithms are highly
data-dependant, it is possible for a compressed data set to be larger
than an uncompressed version.  The notes written to the SAS Log
indicating the effectiveness of the compression should be used during
application development to determine if data set compression offers
any benefits, given the specific structure and content of the data
set(s) involved.

Here is an example of the message written to the Log if compression is
enabled and the data set increases in size:


   NOTE: Compressing data set libref.member increased size by nn.nn
         percent.  Compressed is n pages; un-compressed would require
         n pages.

Operating System and Release Information