Theres A Hole In The Bucket by alendar


More Info
									                           THERE'S A HOLE IN THE BUCKET
There is an article in one of Storage Management Solutions magazine co-authored by Steve
Bentley, Senior Engineer and John Gniewek, Senior Technical Staff member in IBM's Storage
Systems Division titled "A 'hole' New Standard In Tape Standard Reliability," subtitled "Magstar
Stars In ECC Thriller." I found the article to be quite interesting and I would like to share my
thoughts and opinions about this article with you.

The gist of the article is that Steve and John punched holes in a 3590 tape measuring 1.25 mm
which, according to Steve and John, is enough to destroy 100,000 bits of data. They also say that
the punching of holes in the written tape serves as a rough approximation of the effects of debris
that can adhere to the head or recorded surface of the tape. Once again, the source of "debris"
has been identified. I will reference these "holes" later on in this newsletter.

Although they use a 3590 cartridge as a basis for their test, everything they mention or conclude
can be applied to a DLT, Super DLT or an LTO cartridge. The names may change but the
problems remain the same.

They go on to explain that while writing, the automatic read verify feature will detect the presence
of debris and initiate a rewrite in the same place or a short distance "downstream".

The article then describes how the error correction routine works and is able to correct the errors
introduced into the tape in the form of punched holes.

Normally, an optical scanner looks for defects, such as holes in the tape, voids in the oxide,
stretched edges or creases .025" or larger. If you do the math, 1.25 mm equals .04925" or twice
as big as .025". Or said another way, the optical scanner is looking for something that is half the
size of the holes punched by IBM. It would appear, on the surface, that the rejection of a cartridge
that has a .025" hole might be a debatable point with IBM. Regardless, the benefit of the doubt
should always favor the computer.

Incidentally, the size that a defect scanner looks for in a DLT tape or an LTO tape is considerably
larger than .025" or approximately .064" which is a little larger that IBM's .04925" punched hole in
a 3590 cartridge. It is all relative. Bear in mind that a DLT or an LTO cartridge has a much, much
higher density than a 3590 cartridge. We seem to be bracketing them; one smaller and one a little
larger. So, why do physical defect scanners look for certain sized defects? Let's delve into this
and see what we come up with.

Although the error correction routine might be able to correct the situation, there is a rule-of-
thumb that has to be figured into the equation which is, a physically defective tape can never
correct itself. Depending on the size of the defect, there may be sufficient oxide left to pick up the
signal. However, let's apply the rule-of-thumb. A hole in the tape or a void in the oxide can only
get bigger -- not smaller. A stretched edge can only get longer and/or more severe.

Basically what I am saying is that any defect, regardless of the size, is only the beginning which
brings us back to the rule-of-thumb. For example, if you record on a tape with a hole, missing
oxide or a stretched edge, the data could be lost due to the hole or missing oxide getting bigger
and/or the stretched edge getting longer or more severe.

In order to answer the question about the size of a scanner defect, you would have to answer a
question. Would you trust a 3590 cartridge with a void in the oxide .025' or, in the case of a DLT,
Super DLT or an LTO cartridge a defect of .064" or larger? I would also pose the same question
to the systems manufacturers and would certainly expect the answer to be "no".

What if the answer were "yes," which would be very unlikely? I would then ask, "What size defect
would it take to convince you that you have a defective tape?" We are flexible!
This flexibility is particularly true with LTO tapes. The detection method of looking at defects has
changed. Defects can now be measured by length and width and based on the "loss of

IBM in their "Tape and Cartridge Requirements for the IBM Magnetic Tape Cartridge Drives"
under the heading "Tape Discontinuity" says the following:

"Definition: Any physical interruption in the tape such as that produced by tape splicing or

"Requirement: There shall be no discontinuities."

In the same publication under "Inhibitor Tape," the following is stated:

"Definition: Any tape that reduces the performance of the tape driver or other tapes."
"Requirement: No tape shall be an inhibitor tape."

In the list of inhibiting characteristics contributing to excessive errors is "Poor edge conditions".

By definitions and requirements, it would appear that any tape tested by punching holes in them
is out of specification regardless of the outcome.

Why do I include this in an article about holes being punched in the tape? Contamination and
poor edge conditions are not holes. The difference is that contamination and bad edges create
"bumps" as the tape passes over the magnetic head resulting in a "loss of signal" (dropout).
While holes may approximate debris, they are not the only reason for tape failure. "Data
Reliability" might have turned out quite differently had actual contamination been used.

At the end of the article, there is a rather lengthy Editor's Note which warned that this (the
punching of holes) should not be tried at home, which is good advice. I, however, took exception
to one other thing said by the editor.

"The obvious should also be noted: a new cartridge is less likely to suffer from tape errors of the
type discussed in this article and a clean drive is less likely to deposit contaminants onto the
media than one that has been used excessively between cleanings."

The tape deposits contaminants onto the heads which is the exact opposite of what he
said. Although I believe he was describing cross-contamination, he missed the mark. Another
thing is that a magnetic head cannot discriminate between a new (unused) tape and an old
(excessively used) tape. I'm also not sure why he thinks that new tapes are exempt

The article was an extremely good one and no disrespect is intended. I'm merely sharing my
opinions. My compliments to Steve and John.

My experience with magnetic tape also extends to the "punching" of holes in magnetic tape. It's
not as easy as it may seem.

My first encounter was when I was demonstrating a cleaner. The customer wanted to be sure that
the physical defect scanner was, in fact, working. He produced an old tape and said, "Punch a
hole in the tape." In our infinite wisdom we both settled on a paper clip to do the punching. I
straightened out the clip so I had a very sharp end of a piece of metal. The paper clip went right
through the tape and you could see it on both sides of the tape. There was absolutely no question
that there was a hole. Everyone saw it.
Now came my moment of glory! I put the tape on and ran it. Guess what? We didn't find the
punched hole! How could this be???

When I say it's not easy, I mean it. What I actually did with the paper clip is puncture the tape and
made a hole with a "flap." The minute the tape was rewound for the test, the "flap" closed; hence,
no hole and a red face.

To punch a hole in Mylar requires a tool that will cut a circular hole so the center fall out.

All this proves is that the "learning curve" only goes in one direction -- straight up! We never stop

Article contributed by Don Blackburn, Bow Industries, Manassas, Virginia, U.S.A.                 Supplied
August 2003.

To top