on 09-04-201405:58 PM - last edited on 10-10-201509:16 PM by LainieH
Please support an option for much better compression for SAS data sets. This helps reduce (1) disk storage and (2) wait time when the system is I/O bound.
As an example on one data set, even after using COMPRESS=YES I can still get about another 93% compression using 7-zip. COMPRESS=BINARY does not help much.
Enabling the new compression should be as easy as COMPRESS=YES, and reading data sets should require no code changes.
It is OK if this causes the data sets to become read-only and to use more CPU time.
This idea does not include changing SAS data types (like the discussion below, which has gotten off topic).
This can probably be accomplished by working above the record level (so redundancy across records is compressed) and by using better algorithms. Other systems, such as SQLite through the ZIPVFS and CEROD extensions and MySQL InnoDB, appear to do this.