Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Discuss] Backing up LVM partitions using snapshots

On Thu, Dec 15, 2011 at 11:38:58AM -0500, Richard Pieri wrote:
> >That model is changing. The EMCs, NetAPPs and the like don't rely
> >on tape.  They rely on the sort of strategy I am describing.
> >Duplicate reduction and replication, not "backup."
> This is just another tape is dead marketing spiel used to sucker IT
> purchasers.  The promise of "lower costs" convinces them that
> storage frames and cloud storage is the way to go.  It's a trap.  A
> storage frame has finite capacity.  If you want to expand beyond
> that limit then you have to buy another frame or an expansion
> chassis.  Tape libraries are infinitely expandable: you buy more
> tapes.  Those tapes are going to cost an order of magnitude less
> than the cost a new frame.  Really. 300TB of LTO-5 costs
> $10,000-$15,000; a 300TB raw capacity EMC Symmetrix starts around
> $200,000.

These numbers are false though.  If you need to back up 300TB of data
using tape, you need way, way more than 300TB of tape.  You're
probably making multiple copies of the data, to ship one off site, and
you need to muliply that by a factor that reflects your back-up
scheme, and multiply again by your retention policy.  Plus tapes wear
much faster than disks do, so you'll likely need to completely replace
your tapes every 3-6 months.  So, that's going to turn out to be a lot
more than 300TB of tape, on a recurring basis.  You'll probably need
anywhere from 1200TB to 2400TB of tape PER YEAR, at a minimum.  And as
you've admitted yourself, tape is less reliable, so the probability of
data loss is higher given the same number of back-up sets, which is
itself a cost.  

To back up 300TB of data using a scheme like Ed described, you need
300TB of disk space (though you likely still want a second copy to
keep at a DR site, so sure, we'll call it 600TB).  Drives wear, but
much more slowly than tape, and they're probably going to be covered
by some warranty or service plan, so yearly replacement costs will be
kept to a minimum.  And EMC is a top-of-the-line, top-of-the-pricelist
solution; you can probably save a lot of money by using someone
cheaper or possibly building your own systems to do this.  2 LTO-5
tapes (3TB uncompressed) run about $110, a 3TB drive runs about $130.
So how much are you /really/ saving by using tape?  My guess is, over
time, even with the expensive EMC solution, the savings for tape are
negative; substantially more so with cheaper disk solutions.

> I'm not going to say you shouldn't buy disk-based backup systems.  I
> use them myself for some purposes.  They're a useful stage for
> enterprise backups.  They are not the be-all to data integrity.
> Layers of protection is how you maintain the integrity of your data.
> Reliance on any single technology or device is a mistake.

But, it all comes down to cost.  Layering your solution is expensive;
depending on the amount of data you have and its value, it may turn
out that restoring from some old back-up you have laying around, then
abandoning a bunch of lost data you really don't need and recreating
the lost data you absolutely do need from scratch is actually quite a
lot cheaper.  

Back-up schemes can't ever be perfect, but fortunately they rarely
ever need to be.  They just need to be good enough.  The needs of,
say, a large bank are vastly different from those of a small
software shop.

Derek D. Martin   GPG Key ID: 0xDFBEAD02
This message is posted from an invalid address.  Replying to it will result in
undeliverable mail due to spam prevention.  Sorry for the inconvenience.

BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!

Boston Linux & Unix /