Just to add another fly in the ointment, if you have an aggressive
reclamation threshold, say 25%, and a reuse delay of say 5 days, you may
end up with a lot more tapes in a pending state then you anticipated.  A
pending tape is not a scratch tape.

H. Milton Johnson

 
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Coats, Jack
Sent: Wednesday, September 01, 2004 12:42 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: Sizing for a virtual tape library

I agree, think of it as so many tapes.  Since they are 'really fast
tapes', being on disk, you might consider doing reclaimation at some
unusually low number to get expired data out of the way ASAP.  If you
don't need the space that badly, relax the reclaimation percentage a
bit.  Instead of starting withe 80 or 60 percent like you would on
tapes, you might start with 40 percent and go to 20 percent if you need
the space.

This still leaves you with about 20% of the library with 'old data',
plus your need for at least one or two 'scratch tapes' in the library at
a minimum!

If emulating LTO1 drives, at 100G each, 62T gives you about 600 (being
conservative) volumes.  Take away even 2 scratch volumes as a minimum
for scratch tapes and subtract 20% of the rest as 'expired data' you
still get
(598 - 120) 478 volumes at 100G each of real data, or 47.8T on your 62T
library.

If you know your data, and know you get a real world 30% compression,
then it should find close to your 62T of data ( 47.8*1.3=62.14T but that
is close with this level of engineering estimate).  If you get better
compression you really win.

Depending on your needs, you could use client compression, or tape drive
compression.  There are religous camps on both sides, but I suggest you
give it a try both ways to see what you really get.

... green with envy ... JC