 |
 |
Archives of the TeradataForum
Message Posted: Wed, 30 May 2001 @ 16:38:46 GMT
| Subj: | | Re: Compress |
| |
| From: | | Cornelius Meade |
Mark,
This is great information but as in most things there are pros/cons to any issue.... Suffice it to say that in the real world there are
benefits and negatives to any architecture decision.... Where one has the time to perform the necessary analysis of the data beforehand,
then having knowledge of the frequency of values can obviously be used to great effect for fine grained compression. This to me is somewhat
limiting in that you do not always have the luxury of being able to look at the data beforehand or the data may be dynamic and changing. In
these cases a more generic compress capability similar to that found in other RDBMS might be a good option to have. This is a
generalization but with the ever increasing gap between CPU speeds and DASD speeds it would seem that compression features of any type will
be a viable option for the foreseeable future. Spending a few CPU cycles to manipulate the data once you get to the CPU is generally an
acceptable price to pay when getting the data there in the first place is most often the bottleneck. Is there any chance that NCR will
offer a less fine compression option, warts and all?...
| |
