[lug] Tape Drives
gsexton at mhsoftware.com
Mon Oct 28 14:21:24 MST 2002
You obviously did not read the BRU paper did you?
>> And theres no reason you can't accomplish the same with tar. If you're
>> sufficiently paranoid, you can easily do a round-trip archive--
>> extract--compare session with standard GNU tools and (optionally) a bit
>> of shell code. Its quite easy.
If you don't mind the false hits because the files changed between the time
the backup was done and the time the verify was done. I hope you enjoy
looking at log files a lot... I hope that you can exercise the constant
vigilance every day to detect that a failed verify was the result of a tape
problem, and not a user editing the underlying file. Also, I hope it doesn't
matter that you are doubling the load on the disk subsystem because the
verify step will load the disk subsystem to the same extent that a backup
>> So George, do you or a family member work for the Tolis Group? ;-)
That's a cheap shot and I won't stand for it. Not even with the smiley will
I take it. You need to recognize that for some things there are not
equivalent open source packages.
I am a professional system administrator. I can tell you more horror stories
about people who have lost data because of poor procedures, and poor quality
software than you would like to hear. Being cheap and using a bunch of half
baked scripts to save a lousy couple of hundred dollars when you are backing
up data that is worth tens of thousands to millions of dollars is just plain
stupid. When it comes to backup there is no substitute for knowing that it
By all means, save a couple of hundred dollars. But, when you lose data,
your job, or more likely have to send a tape out to a very expensive
recovery house at a cost of many thousands of dollars, ask your self if it
was worth it. I don't know about academia, but in the corporate world if you
are the system administrator, and you lose a lot of very valuable data, you
will be out on the street before you know what happened.
I happen to like BRU because it can verify every file on the tape to ensure
that it wrote correctly. I don't trust tapes, and I don't trust tape drives.
I have been around long enough to know they are not extremely reliable. Even
if you believe that the tape drive is 100% reliable, do you think that the
Linux kernel st driver is totally without flaw? Are you prepared to bet your
job that no one will ever screw up a revision of this driver, or the drivers
for the SCSI sub-system as a whole, or the particular SCSI board? There are
so many layers of hardware and software involved in a tape backup
sub-system, that not checking for correctness on a routine basis is nuts.
MH Software, Inc.
Voice: 303 438 9585
From: lug-admin at lug.boulder.co.us [mailto:lug-admin at lug.boulder.co.us]On
Behalf Of Ed Hill
Sent: 28 October, 2002 1:27 PM
To: Boulder LUG
Subject: RE: [lug] Tape Drives
On Mon, 2002-10-28 at 13:07, Sexton, George wrote:
> Read the paper. You can't check a backup without doing a file compare.
And theres no reason you can't accomplish the same with tar. If you're
sufficiently paranoid, you can easily do a round-trip archive--
extract--compare session with standard GNU tools and (optionally) a bit
of shell code. Its quite easy.
So George, do you or a family member work for the Tolis Group? ;-)
Mind you, I have nothing against commercial apps (use plenty of them on
Linux: MatLAB, LabVIEW, etc.) but on LUG lists I generally point out the
Free alternatives first. Especially in cases such as tar, which in my
experience is one of _the_ most reliable (and cross-platform!) GNU apps
that I've ever used.
ps - If you can't accomplish the same with the GNU apps, I suggest
*you* read up a bit starting with "info tar" and "info md5sum".
Edward H. Hill III, PhD
Post-Doctoral Researcher | Email: ed at eh3.com, ehill at mines.edu
Division of ESE | URLs: http://www.eh3.com
Colorado School of Mines | http://cesep.mines.edu/people/edhill.php
Golden, CO 80401 | Phones: 303-384-2094, 303-273-3483
More information about the LUG