Below are the responses to my question on data storage. Many thanks to all
who took the time to share the experiences with me. The replies are not
edited except the mail headers.
We have decided to get SGI O2. According to the SGI sales rep, the Iomega
Jaz drive is fully supported under IRIX 6.3 and up. So you can simply put
the 1GB cartridge in and click on the icon. That sounds easy. Writeable
CDs have the limitation that they can only be written once. MO drives are
fine if I can get them to work as smoothly, but it costs more than the Jaz
(per MB).
Thanks again.
Jiejun
-------
From: jwu@orion.oac.uci.edu (Jiejun Wu) MY ORIGINAL QUESTION
Dear colleagues,
This is a general question concerning ways for students to routinely backup
their data on SUN and SGI computers. I plan to add a few workstations to
the lab. As I can think of it, there are a few media that can be used:
3.5" floppies, magnetic tapes, magneto optical drives, writeable CDs, and
ZIP and Jazz drives. Obviously floppies have limited capacity and magnetic
tapes are too slow, and they are out of the question. One big concern is
how safe the stored data will be when around the magnets. Ease of use is
another concern. I would like to hear your experiences with any of these
products. I will post a summary afterwards. Thank you.
-------
From: "leonard charles dickinson" <charlie@telemann.pse.umass.edu>
The UMass NMR Facility is a moderate sized operation with 'two' levels of users
with respect to stoarge (actually three, but I won't count those who are happy
one they have a copy on a piece of paper.)
We do no permanent data storage in the Facility and have erase/date policies.
Small time users are happy with floppies: they can pull data thru NMRLINK from
the AC200 or MSL300 and vis GETFILE and File Manager put the files (in WINNMR
format) onto a floppy and have a copy in their own hot little hand. THey can
replot for any purpose. This also works for the minor users of the AMX500, or
ASX300 of DPX300....data can be transferred to either of 3 PCs running WINMR
and then to floppies. This frees up data processing on the INDYs.
The mega-users handle their own data storage on their own INDYs.
We are installing an offline INDY R5000 next week with 4 Gigabytes to allow
more buffer of storage for the users but away from the spectrometers.
Hope this is helpful.
-- leonard charles dickinson|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| | | |Dr. L.Charles Dickinson voice mail and office (413)5771428 | |Director of NMR Instrumentation lab telephone (413)545-0630, -5413 | |Conte National Center for Polymer Research fax (413)5450082 | |Department of Polymer Science and Engineering office Conte Rm 428 | |University of Massachusetts | |Amherst MA 01003 USA charlie@telemann.pse.umass.edu | | |
|"But beneath what the mind choses to admit to itself lie convictions | | that shape our live" _What's Bred in the Bone_ , Robertson Davies | |>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<|
------
Date: Thu, 9 Jan 1997 11:19:02 -0800 (PST) From: "T. Pratum" <pratum@u.washington.edu>
I personally have had good luck with ZIP disks, but haven't used them on the SGI's. I would be interested in hearing from anyone that has. We use them here on a PC and a Mac for backing up data, and I think they have a lifetime which is finite, but probably exceeds the average Chemistry graduate students time in grad school.
Tom Pratum Dept of Chemistry Box 351700 Univ of Washington pratum@u.washington.edu http://weber.u.washington.edu/~pratum
------- Date: Thu, 09 Jan 1997 13:00:19 -0700 From: "Charles L. Mayne" <mayne@atlas.chem.utah.edu>
We have been using recordable CD's for a while now and they are working fine. The major drawback is that the only software I can find that allows easy transfer of data to the CD in small amounts without unacceptable overhead runs only on a Macintosh. Consequently, we have a Mac with a CDR drive attached to our campus network. Users are allowed to store things temporarily on local hard disks on the spectrometer workstation. Then they can ftp the data directly to the Mac CD. The cost is about $8.00 per 650 Mb, and the media are insensitive to magnetic fields.
Charlie
--Charles L. Mayne University of Utah Dept. of Chemistry B103 HEB Salt Lake City, UT 84112 Voice: (801)581-7413 FAX: (801)581-4763 or 8433 E-mail: mayne@chemistry.utah.edu http://www.chem.utah.edu/atlas/facilities/nmr.html
------- From: g-sukenick@mskcc.org (George D Sukenick) Subject: Re: Data backup options for students
Tapes are easiest to store, but not restore; good data density but reliability may be wanting, especially after walking up to the magnet with the cartridge in your pocket.
I'd suggest magnetic optical. If data sets are of reasonable size, you can get 230 MB drives and disks at pretty good prices now. Larger capacity MO drives would be more practical in terms of data storage, if the $$ are present. I haven't had problems hooking them up to SGI.
I've had success hooking up zip disks.
------- From: ALCONFTW!RNDMAIL2!thomasdm@alconatt.attmail.com (Dwaine Thomas)
We routinely backup data from our Varian VXR with tapes and then use an SGI to transfer to a 270MB syquest removable. This is very convenient, space saving, and easy to archive. We have been doing this for three years with no problems yet. The SGI is approximately 13 feet from a 270 MHz wide bore. We are, however, considering to go with re-writable CDROMs once they hit the streets and the price drops.
Dwaine Thomas
Alcon Labs, Inc., M/S: R2-19 6201 South Freeway Fort Worth, Texas 76134
Phone: 817-551-8003 FAX: 817-551-4584 E-mail: ALCONFTW!RNDMAIL2!thomasdm@alconatt.attmail.com
------- From: Jessica Dion <jdion@zoo.uvm.edu>
Syquest drives are another possibility. I have a friend who has had REALLY good luck with ZIP drives - the media is fairly cheap and rewritable, which is a big plus. He bought one for each of his kids at home, so they wouldn't fill the family hard drive up with bunk. I'm not sure how well they work with UNIX hosts. writable CDs are a great way to archive important data, because they have the least probability of being corrupted, particularly around huge magnetic fields. They can only be written to once though, and it wouldn't be worth it for "each student" - I would think it would be more appropriate for periodic archiving of all of the data on the system.
We've recently had bad luck with an 8mm tape (a segment in the middle is corrupted and we'd like to retrieve the data near the end, but TAR quits and won't read beyond that segment of the tape, which became crumpled mysteriously in the drive), and we've been contemplating back-up and archiving alternatives. It will be interesting to read your summary. Good luck! - Jessica Dion Biochemistry Dept. University of Vermont ------- From: Marty Pagel <mpagel@indiana.edu>
Our web page, http://nmrsg1.chem.indiana.edu/data_archives.html, explains the choices that our facility users have when considering data backups for NMR and general Sun/SGI files; it essentially includes the options that you list below. By far the best media for us has been CDROMs and our CD recorder; we also use DAT tapes as a secondary backup, and some individuals use ZIP disks.
Perhaps the best choice of media that you can buy is extra large disk capacities on your new workstations. It is a LOT easier to work with large data capacities if you have an extra 2 or 4 GBytes of scratch space for your few new workstations.
Marty Pagel mpagel@indiana.edu --- __o Department of Chemistry Phone: (812)-855-6492 --- \<, Indiana University Fax: (812)-855-8300 --- ()/ () Bloomington, IN 47405-4001 http://nmrsg1.chem.indiana.edu/~mpagel
------- From: Mike Maddox 5711 <Mike.Maddox@roche.com>
We find that CD-R technology is relatively cheap (drives ~$500 and disks ~$9)and provides a very safe archival of spectroscopic data (650 Mbytes/platter). Multisession capability is there but we prefer to create the entire disk at once.
We also use magneto-optical storage and have not had problems around the magnets. Floppys must be kept at some distance but are ok if your careful.
mike.maddox@roche.com
------ From: "Edward T. Chainani" <etch@ginto.chem.admu.edu.ph>
I backup entire experiment directories using magnetic tapes (4mm DAT type) and indeed they are slow (read write 2 GB in 3 hours). They are good for backing up a lot of data as they can hold 2 GB, but can read in only whole directories. I store the tapes in the NMR lab itself, and have not encountered any problems. Just keep them outside the 5 gauss line, to be sure.
I've used Zip drives with PC's and find that they are faster than floppies but slower than hard disks. Since they store 100MB chunks in a disk, this may be a convenient size for student data. The disks are about $15 each. They're very easy to use with a PC running Windows--check and see if these can work with UNIX workstations as well. I recommend this.
SGI computers usually come with a MO (magneto-optical) drive. I've no experience with this, but it can be a viable option.
------------------------------------------------------------------------------ Edward T. Chainani Nuclear Magnetic Resonance Laboratory Manager
National Chemistry Instrumentation Center (NCIC)
Ateneo de Manila University phone: (632) 924-4601 loc. 2309 Loyola Heights, Quezon City e-mail: etch@ginto.chem.admu.edu.ph Philippines ------------------------------------------------------------------------------
From: allan@KERSHAW.usc.edu
Here at USC, we do not have any SUN or SGI systems associated with the NMR facility, however, the methods that we do use might be implementable on yours. Our campus computing facility maintains a large scale, high speed tape backup system called UNITREE. My facility uses this for archiving all data files. All data files collected on the facility instruments are downloaded to a PC-based server with a 2 GByte disk. They are stored on this disk for 90 days. After 90 days, the data is transfered to the user's UNITREE account. If the user chooses not to have an account, then the data is deleted. All data on both the server and UNITREE is available to the users via an FTP transfer.
We have been using this method for about four months now and it appears to be working well. The data safety is as good as the campus facility can make it. If the users still have concerns, they have the option of backing it up for themselves any way they feel secure with.
As far as speed is concerned, I have not had too much trouble getting my data. The longer the data is on the system the longer it does take to retrieve, but so far the longest time I have waited to get data back is about five minutes.
I hope that this is useful for you. If you would like to discuss this further, give me a call.
Allan Kershaw Director of NMR Instrumentation USC OCW 110 University Park Los Angeles, CA 90089-0744 Office: 213 740-6376 Fax: 213 740-0930 E-mail: allan@kershaw.usc.edu
-----
From: Steve Philson <philson@nmr.chem.umn.edu>
I have used magneto optical drives, but had problems with reliability; now I archive the data myself to writable CD-ROMS, and maintain them centrally for everyone. If you really want people to do their own archiving, the ZIP and Jaz drives look attractive, but I have no experience with them.
-- Steve Philson philson@nmr.chem.umn.edu Director NMR Lab 612-626-0297 Chemistry Dept. University of Minnesota----
From: Eric Shankland <shanklan@u.washington.edu>
we use magneto-optical drives (~220 Mb cap) for most data in current use. for archives we have tapes, including 2 Gb .dat tapes for dumps. by far the most popular is the magento-optical, they are quick, can be remotely mounted, etc. i have no clue how sensitive they are to magnets, though as the nearest magnet to the drive and access to the drive is ~25 or so feet (a 4.7 T horizontal bore with sufficient fringe fields to warrant its own room...)
Eric Shankland Univ of Wash Department of Radiology Seattle, Washington 98195
---------- Jiejun Wu NMR Facility Manager Dept. of Chemistry (714)824-6010 (office) 516 Physical Sciences 1 (714)824-5649 (lab) Univ. of California, Irvine (714)824-8571 (fax) Irvine, CA 92697-2025 jwu@uci.edu (email)