The effect of compression is not linear if you are compressing various types of
data. To illustrate:
00E3A9                        BKTPOOL1        3575DEVC1        36,598.5
100.0       Full
00E390                         BKTPOOL1        3575DEVC1          6,794.6
100.0       Full
Both volumes are Magstar MP C-FOrmat XL 7gb tapes. I would recommend summing the
 "full" volumes and dividing by the
number summed to get an average full volume value which will reflect the various
 data types being compressed. I am
using compression at the drive and not the client.

George Lesho
Storage/System Admin
AFC Enterprises







Paul Zarnowski <vkm AT CORNELLC.CIT.CORNELL DOT EDU> on 02/15/2001 05:36:34 PM

Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>

To:   ADSM-L AT VM.MARIST DOT EDU
cc:    (bcc: George Lesho/Partners/AFC)
Fax to:
Subject:  Re: Formula to calculate # tapes required



At 04:50 PM 2/15/2001 -0500, Prather, Wanda wrote:
>The formula I use is:
>
>Need to back up xxx TB
>Divid by the number of GB that will fit on a tape.
>Adjust as you have suggested for extra versions
>Divide by .65 on the assumption that the not-yet-reclaimed tapes will be, on
>average, 65% full
>Multiply by 2 for the offsite copies

This is pretty close to what I use, too, but make sure you factor in
compression somewhere.  Also, if you are using collocation, then you also
need to consider how rapidly your data is changing.  If you have high data
turnover, then you will be reclaiming data more frequently.  If you don't
have enough tape drives or wall clock time to handle this, you will need
more tapes, plan on a lower average tape utilization, and raise your
reclaim threshold to relieve the pressure on reclaim.

This reclaim issue will not surface at first in a new robot.  You will only
run into it over time, after a bunch of your tapes have filled up and you
start doing more reclaims.  This issue is also dependent on the tape
technology you are using, as some perform reclaims faster than others.

..Paul