Here is a typical amanda report using software compression so do you think 40%
is safe?

HOSTNAME     DISK        L ORIG-KB OUT-KB COMP% MMM:SS  KB/s MMM:SS  KB/s
-------------------------- --------------------------------- ------------
hacksaw      /              0 1146303 454464  39.6   8:40 874.3   0:2816222.2
hacksaw      /var         0   89887  75936  84.5   1:041178.9   0:107392.5
puffer       /                 0  530047 270016  50.9   2:211909.1
0:387160.6
puffer       /opt            0  346710 136224  39.3   7:55 287.0   0:0816364.1

puffer       /u200         0 7352220 821696  11.2  79:27 172.4   1:1311224.6
puffer       /usr            0  855007 306496  35.8   3:171559.3   0:2611951.2

puffer       /var            0  152383 101152  66.4   1:131392.6   0:0911841.4




Joshua Baker-LePain wrote:

> On Tue, 13 Aug 2002 at 1:38pm, Scott Sanders wrote
>
> > OK I know that's a bad thing to say around here BUT...
>
> No, not really.
>
> > ufsdumps (level 0 every night) to tape using amanda. My question is,
> > since the drive is handling the compression what tape length should I be
> > specifying in my tapetype definitions? For example should I use 35000
> > mbytes or 70000 mbytes for a DLT-7000 with 35GB of native capacity? Or
> > maybe something in between just to make sure I don't run out f tape?
>
> Everybody's favorite answer -- it depends.  How compressible is your data?
> Our /home partitions here compress on average about 50% in software.  Our
> raw RF data on the RAID does *not* hardware compress in my AIT1 drive.
>
> Start with some compression estimate based on your data, and lower the
> length if you consistently hit EOT.
>
> --
> Joshua Baker-LePain
> Department of Biomedical Engineering
> Duke University

--
Scott Sanders

Systems Administrator
Concepts Direct, Inc.
2950 Colorful Ave.
Longmont, CO 80504

(303) 682-7110 Phone
(303) 682-7140 Fax