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

Re: [ccp4bb]: Linux tape backup

***  For details on how to be removed from this list visit the  ***
***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***

Richard Gillilan wrote:

> There has been a lot of discussion here about backup
> but no ideal solution. CD-R's are too small, DVD's are
> are interesting, but still not standardized and very
> slow to write I have heard.

First-generation DVD-RAM drives (2.6 gigs per side) write at 1.38
megs/sec, or half that speed with write-verification enabled. 
Second-generation DVD-RAM drives (4.7 gigs per side) write at twice
those rates.

> DLT tapes and the like
> are rather costly. In general, I don't trust magnetic
> media for long term storage and would much perfer
> an optical solution. 

I would be careful about these generalizations.  With CD-R, at least,
there are widely varying reports of the reliability of various
combinations of media types, burners, and readers. See for example:

Some people have found brand X CD-R units work well with media type Y,
while other people with the same unit have had different results.
Recording a disc at 4x may make it unreadable on some drives, even
though a disc recorded at 2x on the same drive works fine.  To top it
all off, someone observed that discs burned with one brand of CD-R
weren't readable in cheap CD-ROM drives, even though the same kind of
media burned in a different device worked fine. The performance of any
piece of media is always a combination of the disc, the drive that
recorded it, and the drive that reads it. A number of specific
discoveries have been posted to Usenet, but none of them are conclusive.
One final comment: while there are clearly defined standards for CD-R
media, there are no such standards for CD and CD-ROM drives -- other
than that they be able to read CDs. It is possible for media to be
within allowed tolerances, but be unreadable by a CD-ROM drive that can
handle pressed discs without trouble. All you can do in this sort of
situation is find a better-quality CD or CD-ROM drive, or switch to a
brand of media whose characteristics are on the other side of the
tolerance zone. 
In general, CD-Rs are far less tolerant of environmental conditions than
pressed CDs, and should be treated with greater care. The easiest way to
make a CD-R unusable is to scratch the top surface. Find a CD-R you
don't want anymore, and try to scratch the top (label side) with your
fingernail, a ballpoint pen, a paper clip, and anything
else you have handy. The results may surprise you. 

This would especially be a concern if your 6 gig data set is on 10 CDs
and a slight scratch on any of the 10 CDs could render the whole data
set useless, or if you have chosen cheap media.  Have you gone back and
checked current CDs that you burned say a year ago to see how reliable
they are?  Since you are currently burning CDs on a Mac, you could check
them with CDRScanner, which can report on weak sectors on the CD:

More on this issue here:

Test results clearly indicated that all discs were not alike, even if
their colors were similar. Cost pressures have resulted in a broad
matrix of stampers, dyes, metallizations, and processes. No correlation
was observed between CD-R quality and dye type (cyanine or
phthalocyanine), metallization (gold or silver), or recording speed
(2X-8X). Quality is primarily determined by efforts at the manufacturing
facility, and depends less on types of dyes or metallizations.

Quantum claims (I don't know how true this is) that DLT tapes actually
become more reliable with increased use.  DLTs are meant for long term
storage, much more reliable than other types of tape, and personally I
would trust those tapes as much as the average CD-R.  I've burned CD-RWs
that have developed read troubles in a few months, just sitting in my
desk (although I believe CD-RW is generally believed to be less reliable
than CD-R). 

You can get the lower-end 20 gig (uncompressed) DLT drives for about
$1500 these days.  If you want rewritability, DLT media is cheaper than
DVD-RAM and similar to CD-RW, when the DLT tapes are used in a 20-gig
drive (in packs of 10 you can get the tapes for $65 each, possibly less
if you shop around).  If you use the tapes in a 40-gig drive, DLT is
cheaper than both--the same $65 tape holds twice as much--but you have
to pay $3-$4k for the higher capacity drive.  At 40 gigs per tape, the
media cost even approaches that of CD-R and perhaps even beats it
depending on how often you end up burning coasters.  It also beats CD-R
costs if you can take advantage of reusing the tapes.  The high-end DLT
drives also have an uncompressed data transfer rate of 6 megs per
second, which will handily beat CD-R and DVD-RAM.  But if your data
compressed well you should do better than that.  We have an older
DLT-4000 series drive (20 gigs uncompressed).  These are supposed to do
around 3 megs per second, but as we are only on a 10-baseT network our
backup speeds are network-limited so I can't comment on the maximum rate
for such a drive with compressable x-ray data.

You might find the following Quantum propaganda interesting since
includes a claim that CESR folks have been very happy with DLT in the

> In general, I don't trust magnetic
> media for long term storage and would much perfer
> an optical solution. Orb drives are very interesting:
> 2.5GB/disk for  30 USD. They claim 12MB/sec
> transfer rate over fastest SCSI.

But they too are using magnetic media.  My experience has been that
removable drives using rigid magnetic platters are  prone to media

I know you said DLT was expensive, but in fact DLT is much cheaper than
Orb if you are going to buy any significant number of tapes/catridges,
which you apparently plan to do.  If you are willing to spend that much
on Orb disks, then you should also look carefully at the current price
of DLT.

Eric Bennett ( ericb@pobox.com ; http://www.pobox.com/~ericb )

The question is not whether machines think, but whether men do. 
-B.F. Skinner