Amanda-Users

Re: tapetype question

2006-08-02 13:49:21
Subject: Re: tapetype question
From: Jon LaBadie <jon AT jgcomp DOT com>
To: amanda-users AT amanda DOT org
Date: Wed, 2 Aug 2006 13:42:52 -0400
On Wed, Aug 02, 2006 at 01:05:33PM -0400, McGraw, Robert P. wrote:
> 
> > On 2006-08-02 15:08, McGraw, Robert P. wrote:
> > > ./sbin/amtapetype -o -t LTO2HWC -e 200g -f /dev/rmt/1n
> > >
> > > Writing 2048 Mbyte   compresseable data:  38 sec
> > > Writing 2048 Mbyte uncompresseable data:  76 sec
> > > WARNING: Tape drive has hardware compression enabled
> > > Estimated time to write 2 * 204800 Mbyte: 15200 sec = 4 h 13 min
> > > wrote 6422528 32Kb blocks in 98 files in 7358 seconds (short write)
> > > wrote 6455296 32Kb blocks in 197 files in 7737 seconds (short write)
> > > define tapetype LTO2HWC {
> > >     comment "just produced by tapetype prog (hardware compression on)"
> > >     length 201216 mbytes
> > >     filemark 0 kbytes
> > >     speed 27315 kps
> > > }
> > >
> > > I am in the process of trying to find out how to turn off hardware
> > > compressing for the Solaris 10 OS.
> > 
> > 
> > I hope you did understand that turning off hardware compression for
> > LTO tape drives is not so important as for most other tape drives.
> [McGraw, Robert P.] 
> Yes I did. 
> > 
> 
> I am planning to leave it as is. I do not software compress at this time.
> 
> Why take up the CPU cycles to software compress your data if it is going to
> be compressed on hardware?


Others may have different thoughts, but here goes.  (btw I use sw
compression, but I'm not a "you really should too" advocate)

gzip, and certainly bzip2, seem to get a higher degree of compression
than most tape drive's compressors.

It matters little after the planning stage except if you are using
a tapealgo like "largest fit".  I think that is the only time
tape length is used after planning.  Once a dump is chosen for
taping, it is sent to the current tape even if unlikely to fit.
So the remaining tape available is unimportant except for that
one algorithm.

During planning some of these values are used.  Tape length and the
size of the dumps may affect the decision to promote or bump.  This
is because amanda will not schedule a set of dumps larger than the
total tape space available (tape length x runtapes).

The planner takes into account the estimated size of the dumps as
the taper will see them.  If using sw compression this matches how
much tape it will actually consume.  If using hw compression, it is
the original size of the dump and the amount of tape it actually takes
after hw compression is a guess (but not one amanda makes).

A situation where this might make a large difference is when you have
vastly different compressibilities of one DLE vs another.  When using
sw compression amanda has a history of the compressibility to use in
its calculations.  Not so with hw compression.

Opting to use hw compression is a fine choice.  It has many advantages.
sw compression has a few also.  If you choose sw compression for every
thing, use the tape length obtained from amtapetype.  If choosing hw
compression (for all or some DLE) increase the tape length value.  How
much to increase it is a guess for you to make.  Then monitor the dumps
to adjust it if problems arise.  Ex. frequent running out of tape says
too high, not being able to dump some level 0s due to apparent lack of
tape might say too low (or a need to bump runtapes).

-- 
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)

<Prev in Thread] Current Thread [Next in Thread>