Re: Disk space problems

Steve Philson (philson@nmr.chem.umn.edu)
Thu, 15 Aug 1996 10:46:27 -0500 (CDT)

There has been some discussion of CD-R, MO, and Zip/Jazz drives, and I
have a few cents to contribute too. Our earliest archiving methods
were with a DAT drive, and when that proved unreliable and difficult
for users, we went to 3.5" MO drives (128 or 230 MB), which we used
for several years on both Unix and Mac systems. We had several
failures of the MOs on both types of system. My theory is that these
drives generate more heat than the people who bundle them into
cabinets realize, and the constant heat causes failures. The real
problem is that when they fail, they tend to destroy the archived data
you are supposed to protect. This may be partly from deficiencies in
the OS or drivers, but the fact that they are writable makes this
always a possibility.

The Zip and Jazz drives are very attractive, but they suffer from the
same danger as above, they are always writable and therefore
corruptible. Besides, they are strictly magnetic, and may be used in
an environment with large stray fields. Probably nothing large enough
to be a problem, but you have to stay alert. Still, they clearly
provide a convenient storage and transfer method for individuals.

We began some months ago archiving our data to CDs, using a
CD-recorder and some free software for our Solaris systems. We no
longer ask individuals to back up their own data; every couple of
months I dump the oldest data and write it to a CD. The most recent 5
of them are on line always, and any other one can be loaded on
request. This has been mostly satisfactory, and if I start getting
too many requests for old data, I will simply buy a few more CD
players. This system is more trouble for me, but less for the users,
and a lot less prone to disaster (I hope!!).

Every lab has different requirements and different expectations from
users. We try to make it difficult for users to make errors, and
archiving data is certainly a place where mistakes are both easy and
costly.

--
Steve Philson           philson@nmr.chem.umn.edu
Director NMR Lab        612-626-0297
Chemistry Dept.         University of Minnesota