Workable copy pools is *possible*, provided all primary storage pool is on
disk; that is, all backups go to disk pool, which is sufficiently large to
hold all the versions for retention... and never does migration!
I've had to do this (and more) for a capital equipment-constrained customer;
12 locations around the world, getting the "same" equipment (from Dell) for
service replacement purposes. Most sites had two drives (one in each 120T),
one site had two drives in a single library (130T), but TWO SITES WITH
SINGLE DRIVE LIBRARY needing onsite backups got 105 GB disk pool (sufficient
to provide 14-day point-in-time restore for approx. 55-65 GB, depends on
daily turnover)... so, with six usable slots, we configured 2-slots for
onsite copypool, 2-slots for offsite copypool, 2 slots for db-backups.
Reclamation was done as if both copy pools were offsite, so fresh tapes
would be cut from the primary disk pool -- so, single drive reclamation was
not required. For the 1-drive, 2-library sites, single drive reclamation was
done AND it worked just fine (using 100 GB disk pool, scheduled for weekends
after disk migration)... all these sites ran fine for over two years, the
only true "glitches" were due to onsite tape management needing occasional
assistance with their DRM actions. Also, turns out, we rarely had tape
drive failure after the initial install/burn-in period -- even then, we had
less than 5 drives fail across all 12 sites.
NOT the *best* answers, but these configurations did allow us a lights-out
environment --- biggest caveat is when (not if) the drive goes down, no
backups to tape (slightly mitigated by using LAN to store db-backups off to
another server).
Yes, running with less than 3 drives is a challenge, but a lights-out setup
*can* be done with only 1 drive PLUS a large disk pool for the primary
storage pool!
(My motto, had to be: "If you bring money, we can solve"!!!)
Don France
Technical Architect -- Tivoli Certified Consultant
Tivoli Storage Manager, WinNT/2K, AIX/Unix, OS/390
San Jose, Ca
(408) 257-3037
mailto:don_france AT att DOT net
Professional Association of Contract Employees
(P.A.C.E. -- www.pacepros.com)
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU]On Behalf Of
Zlatko Krastev
Sent: Saturday, August 17, 2002 8:09 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: 2 drives are required for LTO?
Mark,
I fully agree with your opinion. TSM *can* work with single drive but it
would be ugly. Same waste of resources as assignment to 5k project a
project manager with 300k salary (and you can always send 1 kg parcel with
a truck).
I said it is possible but will never say I recommend it.
Zlatko Krastev
IT Consultant
Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
cc:
Subject: Re: 2 drives are required for LTO?
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU]On Behalf Of
Zlatko Krastev
> The requirement for 2 drives is not mandatory, it ought to be just a
> suggestion. LTO can be used as a standalone drive, TSM can use single
> drive or small autoloader with single drive. So it is NOT required but
> recommended.
Yes, technically a single drive is sufficient to do backups. But then I
could use a pair of nail scissors to mow my lawn...
> - single drive reclamation - define reclamation storage pool of type
FILE.
> On reclamation remaining data is moved to files and later written to new
> tape volume. Drawback: data is not read when written (sequential
> read+write vs. parallel) thus takes more time. Calculate time budget
> around the clock.
FILE storage pool-based reclamation is dog slow, and expensive of disk
space, particularly if you are backing up database-type data of any size.
I've got a customer trying to do this very thing, and reclamation is
extremely slow.
> - single drive copypools - define following hierarchy DISK -> FILE ->
LTO
> (file pool would be also lto reclamation pool). Prevent file->lto
> migration during backups (highmig=100). Perform backups to copypool
after
> node backups finish. Allow migration after backup to copypool finishes.
> Drawbacks: filepool must be large enough to hold all backups data.
Backups
> should not happen during migration because some object(s) may migrate
> without being copied to the copypool. Again time - data have to be
written
> twice through the one-and-only drive. And on the end with one drive
there
> is no way to perform copypool reclamation.
Bingo. A single tape drive, because of the lack of reclamation, means no
usable copy pool, no way to use move data to consolidate primary tape
volumes, and no way to use a restore volume command to rebuild bad primary
pool media from copy pool media--in short, a badly crippled TSM backup
system.
> Conclusion: for a small installation data might be not too much, time
> might be enough for all activities (node backups, copypool backup,
primary
> pool raclamation, migration, DB backup). Thus neither LTO technology nor
> TSM dictate number of drives to be used but only the business
requirements
> you have.
Don't let anyone tell you that a single tape drive is adequate for
anything
resembling a real backup system. If you can't afford a real library, you
can't afford a real backup system. My experience with multiple
environments
calls for a minimum of three drives--two drives for multitape operations,
and a spare in case one drive breaks down.
--
Mark Stapleton (stapleton AT berbee DOT com)
Certified TSM consultant
Certified AIX system engineer
MCSE
|
