[AV Media Matters] Volumetric Density and Tape Thickness for Magnetic Tape Syste

[e.ha.zwaneveld@nfb.ca]

Greetings,

Jose Llufrio of ICAIC asked today about data track and volumetric density
for various magnetic tape recording systems in relation to the discussion
about robustness of DV tapes:

The 1998 National Storage Industry Consortium LINEAR TRACK DENSITY
perspective was as follows:

LINEAR TRACK DENSITY:
In 1998, linear track density was 750 TPI and for helical 2800 TPI; in the
year 2002 these numbers are projected to be 6000 TPI for both linear and
helical systems; in 2007, they are projected to be 20,000 TPI for both
linear and helical magnetic recording systems. So the higher density of the
DV format is a trend that is a fact of life. When I discussed it with a
Panasonic scientist recently, he pointed out that the error correction and
robustness of the DV format is actually superior than the older
alternatives. Obviously, we should expect the equipment to warn us when the
error rate exceeds the ability of the error-correction circuit to do its
work and when error concealment sets in which if permitted to go on without
migration of content, means that you have missed your chance of preserving
content. So robustness no longer means 'THICK TAPE' but a 'ROBUST SIGNAL'
which signals when its integrity is threatened and expects this to be acted
upon in a timely manner. In my opinion this means that automated storage and
maintenance systems are the only affordable way to archive magnetic media
content.

There were also expressions of nervousness in the earlier discussion about
the Volumetric Density in Magnetic Tape Systems. When we review the
Technology Roadmap, also prepared in 1998 by the National Storage Industry
Consortium, we note that the trends are toward a REDUCTION in tape thickness
by a factor of 2x in the next 5 years:

High End Metal Particle (MP) tape:
Substrate material: in 1997 PET/PEN/PA; in 2002 it will be PEN/PA; and in
2007 it will be PA or better.
Substrate Thickness in 1997 was 14 microns; in 2002 it will be 5-6 microns;
and in 2007 it will be 4-5 microns.
The Total Thickness in 1997 was 18 microns; in 2002 will be 8-9 microns; and
in 2007 will be 6-7 microns.

Mid to Low Range Metal Particle (MP) tape:
Substrate Material: in 1997 PET/PEN/PA; in 2001 it will be PA and in 2007 it
will be PA or better.
Substrate Thickness in 1997 was 4.5 microns; in 2002 it will be 3-4 microns:
and in 2007 it will be 2-3 microns.
The Total Thickness in 1997 was 7 microns; in 2002 it will be 4-5 microns;
and in 2007 it will be 3-4 microns.

Mid-Low Range Metal Evaporated (ME):
Substrate Material: in 1997 it was PET; in 2002 it will be PA; and in 2007
it will be PA or better.
Substrate Thickness in 1997 was 6.3 microns; in 2002 it will be 3.5-4.5
microns; and in 2007 it will be 2.5 microns.
Total Thickness in 1997 was 8 microns; in 2002 will be 4-5 microns; and in
2007 will be 3 microns.

The subsequent reduction rate will be slower. If I were in the tape
restoration business though, I would see this trend as a potential
business threat. One expert has already expressed doubts about his
ability to 'restore' tape thinner than 10 microns. Archivists and
other users of magnetic tape will as Asset Managers develop
preventive methods to assure signal integrity for as long as desired
and justified. So we either invent new content recovery methods from
future degraded tape, or expect a decline in the amount of
restoration business.

Those of you who attended the AMIA Preservation Committee meetings
last year, will recall my discussion of the Linear Tape Open (LTO)
data storage solution.  To conclude, I will provide you below with
some AREAL DENSITY and VOLUMETRIC DENSITY Projections from IBM that
illustrate that this multi-vendor supported product could in the not
so far future provide the higher storage density and data rates
required for video data:

AREAL DENSITY PROJECTIONS:
In 1975, 3420 data tape had an Areal Density of 0.0001 Gb/Square
inch; by 1984, type 3480 in 1984 was 0.001 Gb/si; in 1990, type
3490E was in 1990 better than 0.001 Gb/si; in 1995 type 3590 was
better than 0.01 Gb/si; in 1996, type 3570  was 0.05 Gb/si; and LTO
starts later this year with 100 Mb/si and approaches 1 Gb/si.

VOLUMETRIC DENSITY PROJECTIONS:
In 1975, type 3420 was 0.07 Gb/cubic inch; type 3480 in 1984 was 1
Gb/ci; type 3490E was in 1990 at 5 Gb/ci; 3590 was in 1995 50 Gb/ci;
and type 3570 was in 1996 100 Gb/ci; and LTO is in 1999 500 Gb/ci
and will in 2005 reach 5000 Gb/ci.

The meaning of all this is that we should become concerned about
SYSTEM ROBUSTNESS  and PREVENTION OF SIGNAL INTEGRITY DEGRADATION
given that the much higher densities of digital data on much thinner
tape require intelligent and constant signal monitoring and timely
migration. Old concepts of manual tape evaluation for degradation,
rewinding tape one at a time, evaluating quality of a duplicate by
comparing it subjectively or with the previous generation, without
reference to the objective quality reference criteria recorded about
the approved source will not work any longer.

I hope that this elaboration is useful.

Best regards,

Ed H. Zwaneveld,
Technological Research and Development,
National Film Board of Canada
125 Rue Houde, T-3
Saint Laurent, QC, Canada H4N 2J3
Tel: +1-514-283-9143
Fax:+1-514-283-0278