On Tue, Oct 07, 2003 at 04:51:30PM +0200, Paul Bijnens wrote:
> Bradley Glonka wrote:
> 
> >This tape drive is supposed to have some "intelliget"
> >Sensors to determine if it should turn compression on or Off.  
> 
> The main reason why hardware compression should be off,
> is because most tapedrives blindly apply the compression
> algorithm to any data.  It is a know fact that this INcreases
> already compressed data.
> 
> However, recently we found out (by using amtapetype)
> that some recent LTO drives have some more intelligence
> built into them, and detect when or when not to do hardware
> compression.  The result is that even if you feed it already
> compressed data, the capacity on tape is optimally used,
> even with hardware compression.
> 
> You could even argue that in this case you better have hardware
> compression on, so that you can have some DLE using software
> compression (to lower the network bandwith) and mix with some
> other DLE's without software compression (to avoid the cpu-load).
> You can have the cake and eat it too.


I might agree if I understood the following better.

At what points in a amdump run, or amflush run,
does "tape remaining" become a consideration?

Tape remaining doesn't seem to be a consideration
before starting to tape a DLE (neither holding disk
file nor direct to tape dump).  Amanda will start
taping an DLE that it can easily determine will not
fit on the tape remaining.

With hw compression on, even with "intelligent" drives,
amanda has no idea how much tape remains.  If there are
some points where tape remaining is a significant consideration,
then it might still be better to turn hw compression off.

-- 
Jon H. LaBadie                  jon AT jgcomp DOT com
 JG Computing
 4455 Province Line Road        (609) 252-0159
 Princeton, NJ  08540-4322      (609) 683-7220 (fax)