Re: [rdiff-backup-users] tape drives

[Greg Freemyer]

On 5/26/05, Keith Edmunds <address@hidden> wrote:
> This is an interesting, if off-topic, discussion.

Actually, I'm in the middle of putting together a powerpoint on "data
retention, preservation, and spoliation"  (Spoliation is a legal
phrase for degradation.)

I'll be presenting it at the Alabama State Bar association meeting this summer.
 
> Greg Freemyer wrote:
> > USA law requires archival backup of lots of different document types
> > (including e-mail) for various time periods.  Often years.
> 
> [slightly-sarcastic] Does it require that the backup be readable?
> 
Obviously that is the point, but I'm not sure that penalties would
apply if you show a true attempt.

> I'm sure I'm not the only one who has witnessed someone trying to read
> an old backup tape unsuccessfully. 

Very true.  But with DLT/LTO class tapes, things should be much more
reliable than with DAT, or other low end media.

One of the main reasons is the lack of an alignment track on low-cost media.

Thus, if you drop a dat tape, there is a very good chance it will be
unreadable.  Media that has an alignment track is far more tolerant of
physical shocks.  (I won't talk about fire/heat.)

Good tape media is supposed to have a 20-30 year life.

The longest rated life is 100 years, but that is for microfiche!!!!!! 
Sort of hard to keep a few TB of data on that.

> I also wonder about media - tape
> backups may be on 9track 1/2" tape recorded at 800, 1600, or 6250 bpi
> (and probably other values too), or maybe on DAT or CompacTape or ...
> Can you still find a tape drive to read your data of 20 years ago? 

The government typically only requires 7 years max.  Popular
technologies are usually well supported for at least that long.

During some court cases, older backups are relevant and attempts to
restore need to be made if feasible.

We use specialist companies when the need arises.  One example is Emag
(http://www.emaglink.com/).  They claim to have 1000 tape drives for
doing tape recoveries.  Lots of duplication in drive types, but also a
large selection of drives to choose from.

We sent out a nine-track tape last year, I don't remember who we used.

The costs are high, but it is doable in many cases.

> What
> about today's data in 2025? Of course disks aren't immune to changes
> over time, but an IDE disk of 20 years ago should still work in today's PCs.

How about a MFM drive.  I've forgotten when those where last
manufactured, but it seems they were the disk technology of preference
in the early 80's.  I think they were superseeded by RLL drives for a
few years, then finally IDE.  Can a current PC read those?  (I truly
don't remember if IDE is a superset of MFM, or if it was a new
technology.)
 
Or you can have problems in the opposite direction.  In my profession
it is common to move disks around between computers.  IDE drives over
128 GiB (137 GB) require LBA-48 capability in the disk controller
(IIRC optional in the ATA-7 spec. and not in earlier specs. at all.)
and LBA-48 support in the OS disk driver.  I have seen at least 3
occurances of a user putting a large drive on a non LBA-48 capable
controller.  It sort of works, but when you try to read/write data
above 128 GiB, it actually wraps and you get the wrong disk data.  The
end result is typically massive disk corruption.  Backups become very
important once again.  (In one of those cases, I was being paid to
observe and render an opinion as to the expertise of the individual. 
Obviously, this did not reflect very well on him.)

> Not meaning to attack you, Greg, but my point is that there is more to
> long-term backups than dumping the data on a tape and storing it in the
> attic.

Totally agreed.  

But even worse is dumping data to CD or disk, then storing that in the attic.

Cheap CD's can start to fail after a few months.  The glue on CD
labels can be corrosive. etc. etc.

Disk technology depends on filtering the air to keep the internal air
clean.  Powered off the internal lubricants pollute the air
(disrupting the magnetic field) and can settle on the platter and/or
the head.

I recently had to work with a drive that had been powered off for two
years.  Upon powering it on, I got CRC errors for about the first 30
minutes.  After that, it began to work consistently.  My assumption is
that after 30 minutes the air was filtered enough that the magnetic
field could consistently make it across the air gap between the
platter and the disk head.

Overall, from what I have read and the high-end industry symposiums I
have been to the recommendation is still to use high-quality tape for
long term archival purposes.

> 
> --
> Keith Edmunds

Greg
-- 
Greg Freemyer
The Norcross Group
Forensics for the 21st Century