Every time my fancy
modern computer crashes, Im reminded of a scene from a Star Trek episode circa 1990.
Captain Picard, finding himself in some life-and-death pickle or other, cleverly issues an
S.O.S. by inputting computer code into his android crewmate, Data. That message, a humble
binary command, is later found by the crew so the show can end on time.
If only it were so easy to raise a crashed Mac. But the more profound message here lies
in the subtext: that 24th-century computer programming will be based just as it is
now on binary mathematics.
So, is that just a bunch of Hollywood hooey crafted to sell Excedrin commercials?
Hardly. A dozen years after showbiz writers posited this theory and 63 years after the
first electronic computer was invented, binary still rules. And thats the beauty, as
well as the frustration, perhaps, of digital data. The foundation never changes, which is
kind of a relief, but the recording media well, thats another story.
Connect the dots
Lets start with binary. First of all, what is it? Quite simply, it is
interminable strings of ones and zeros, and it is the only language computers actually
understand. Its also known as machine language. Binarys ones and zeros form
digital bit streams that trigger on/off (or up/down) electronic signals that ultimately
drive computer functions. The one and zero bits form complex codes that are managed with
assemblers, or high-level commands, so that the machine language is transparent to
programmers and easier to deal with (although programmers may certainly choose to write
code in binary, if they dont have a life). Programs are merely binary code
compilers.
Every electronic file regardless of origin or how it may be humanly manipulated
at higher programming levels remains a binary file at its core. Taken all apart,
then, every digital image consists of blocks of ones and zeros; these are written as
raised spots and pits (on optical media) or magnetized and non-magnetized spots (on
magnetic media). So whether you store it on a portable medium, 200 GB hard drive,
microchip or, uh, humanoid subsystem, the file itself stays the same: binary. Thus was
Capt. Picard able to save the universe, and the increasingly computerized world of
healthcare is assured a digital file continuum.
The very first computer to run on binary was the Atanasoff-Berry Computer, or ABC,
built in 1939-1942 by Iowa State University professor John Atanasoff and graduate student
Clifford Berry. At 64 wide, 45 high, 33 wide, and 800 pounds, it
was also the first electronic computer. ABC had 3,000 bits of memory in two banks the size
of large coffee cans and could perform one operation every 15 seconds. (Modern computers
perform 10 billion operations per second.) It was built from telephone switchboards, 300
vacuum tubes, and a mile of wire. The recording medium was electronically punched cards.
Since then, recording media have evolved a lot, from paper, wax and vinyl to film,
magnetic and optical. So while digital content structure remains virtually unchanged,
storage for it is anything but.
MOD, MOD world
The type of medium you choose for archiving medical data depends on many
variables: what youre saving, how big it is, whether it will be altered, when
youll need to access it again, and how much automation you require.
Large facilities and storage service providers started archiving to tape or DVD for
reasons of storage capacity, but magneto-optical disc (MOD) technology has found many fans
in healthcare. Unlike tapes and hard drives, MO discs are resistant to magnetic fields,
which is what initially got them into radiology systems.
All of the CT and MR machines that were shipped throughout the 90s had MO
drives on board. Thats why it kind of caught fire, says Mike Sutherland,
medical channel manager for TDK Electronics Corp. (Garden City, N.Y.). The problem
was the storage capacity. Thats why the MO group has raced to ship its 14X product,
which is 8.6 or 9.1 GB. Some of the large manufacturers have jumped on that now for
archives.
MOD combines laser and magnetic technologies, meaning the discs are written
thermo-magnetically and read optically. The new 9.1-GB 5.25 discs have twice the
capacity of DVDs. Verbatims (Charlotte, N.C.) have an advertised archival life of 40
years, and Hewlett-Packards (HP of Palo Alto, Calif.) claim 100 years.
MODs come in both write once/read many (WORM) and rewritable versions. Like other
removable discs, MOD as an archive is limited to the capacity of its largest jukebox,
which gives only about 2.2 TB of storage using 9.1 GB discs. But thats fine for
nearline hospital storage and long-term storage for smaller facilities.
Rose Honea, R.T.R., R.D.M.S., is the PACS manager at Texas Childrens Hospital in
Houston. She says TCH has been using a Philips [Bothell, Wash.] MOD archive since
implementing its IMPAX picture archiving and communications system (or PACS, from Agfa
HealthCare, Ridgefield Park, N.J.) in 1991. It seems to be pretty reliable.
Were about 99 percent trouble-free. For redundancy, we also copy all our exams over
to [StorageTek] 9840 tape.
She says they do encounter the occasional blown MO disc. You could copy them
twice to MOD, but so many exams fit on that medium that once one is corrupted,
theres no guarantee that all exams are retrievable.
TCH logs 147,000 exams per year, all of them pediatric. By Texas law, they must be kept
for 22 years. The hospitals network has a powerful 100 base-T gigabit backbone;
Honea says it enables PACS users to retrieve both MOD and tape images in about a
minute and a half. If the exam is older than six months, its going to take maybe
three minutes. Overall, she says, Its working out fine, although
the staggering volume of images the hospital must keep has caused it to consider
outsourcing its archival storage.
Tape it to the limit
Among the least stable media of all time thats what the
Council on Library and Information Resources (CLIR, Washington, D.C.) unapologetically
called magnetic tape in a paper it published in the late 1990s, Into the Future: On the
Preservation of Knowledge in the Electronic Age. Digital technology
is not
ideal for preservation purposes, deadpanned CLIR when noting the fury of NASAs
Jet Propulsion Laboratory (Pasadena, Calif.) upon discovering that 20 percent of vital
data taped from the 1976 Viking Mars mission were corrupted.
Nevertheless, tape is still widely used in medical archiving, its primary attraction
being capacity. There are linear technologies such as LTO and DLT, and helical varieties
such as DAT, AIT and DTF. Linear records data on tape in a straight line. Helical records
data in diagonal stripes across the tape.
One company that swears by linear is StorageTek (Louisville, Co.). Howard Hayakawa,
V.P. of StorageTeks tape drive product line, says, We sold a helical-based
tape drive [Redwood] a number of years back and it was very successful, but it was a very
difficult system for us to support. It was introduced in 1992; the transports
manufacturer stopped production, and support has since been plagued by unavailability of
parts. StorageTek now sells DLT, LTO and its own half-inch tape designs, 9840 and 9940.
Hayakawa says the 9840 has two reels inside so the tape never really leaves the
cartridge; you dont waste time threading tape. And because the tape is all inside
the cartridge, weve put half the media in one and half on the other, so you start in
the middle of the cartridge. It takes far less time to get to the end. It allows you an
average time to data of 12 seconds. The 9940, by comparison, takes 58 seconds, while
LTO and DLT take more like a minute and a half. If youre looking for a 25-MB
image on the 9840, you can get there in about 13 seconds. For people who really care about
[speed], there really is no other solution except [hard] disk, which tends to cost
more.
He says StorageTek customers typically want 9840 for nearline storage, and
midrange drives like the LTO for archiving, because there youre not quite as
concerned about how quickly you get the data back.
Rorke Data, Inc. (San Jose, Calif.) also chose tape for its RTL archive line,
configured with either AIT (the Sony/Matsushita product) or LTO (developed by HP, IBM in
Armonk, N.Y., and Seagate Removable Storage Solutions in Scotts Valley, Calif.). LTO
stores three-fourths the volume of AIT-3, but its data transfer rate is 27 percent faster
(115 GB/hr vs. 84 GB/hr). AIT, LTO and 9840 all claim archival life of 30 years. Rorke
also sells archives with DVD and CD.
Weve seen DVD more prevalent in cardiology, and tape more prevalent on the
radiology side, says Aaron Blotsky, regional sales manager of Rorkes medical
imaging applications. He says each has merits. AIT-3, for cost per MB, is still the
best archiving medium available today. One AIT tape holds 100 GB of native data or
260 GB compressed. LTO stores 100 GB native/200 GB compressed. DVD only holds 4.75 GB but
gives quicker access, and DVD libraries are relatively inexpensive and are pretty
similar [to tape] until you get to the 5 TB range due to limitations of jukebox
size.
The only problem with todays tape systems, Blotsky maintains, is that they are
not really an enterprise-class type product. Because of that, Rorkes
next archive product is based on an entirely new product, Super AIT.
Despite the name, Super AIT from Sony Electronics Inc. (Park Ridge, N.J.), or S-AIT, is
not another version of the 8mm helical series. Essentially its a longer, wider
(half-inch) AIT.
The question came up: Were able to pack so much information onto
small, thinner tape [AIT], what if we made the tape as big as DLT or LTO how much
would it hold then?, says Kevin Handerson, Sonys director of data
technologies.
The answer: 500 GB uncompressed. Customers who want to archive a large amount of
information are looking for a bigger bucket. Its a little bit slower than AIT, but
its as fast as any linear tape out there and holds much more. Because the S-AIT
cartridge has a single spool, it does have to fully rewind before its unloaded or
loaded into a machine. But the search time is less than a minute, and then its just
a few seconds to bring the file up.
Bigger, faster, roomier. Is there a downside? Only that S-AIT is a proprietary format
controlled by one company that aggressively upgrades product. Yes, youll need a new
drive. (So much for backward compatibility.) For the gun-shy, there are always options.
We dont have any proprietary hardware or software, says Samad Moini,
senior executive for storage life sciences solutions at IBM. Tivoli [an IT
management program] is an IBM product, but it runs on Sun, HP, SGI, Windows, and Linux.
Its open. If a customer does not want to use Tivoli, they can use Veritas or Legato
[both in Mountain View, Calif.] with our LTO solution and do the same archive.
In April, IBM succeeded in shoehorning 1 TB of native data onto one of its Enterprise
3590 cartridges, making it the highest-capacity tape extant. It was constructed with
advanced material made of super-small particles from FujiFilm (Stamford, Conn.). The drive
technology already exists, and the tapes will be commercially available in 4 to 6 years
just about when Sony will have a drive capable of writing 10 TB of data to S-AIT,
which the company claims that tape can already hold.
DVD
Does anyone remember LaserDisc, the wanna-be successor of VHS from the 1970s?
Does anyone want to?
As big as LDs were (an ungainly 8 or 12 inches), they still had to be flipped over in
the middle of a movie because the whole thing wouldnt fit on one side. And as
everyone knows now (especially those whose LD enjoyment of Star Trek IV was impacted by
the Leonard Nimoy interview that could only be avoided by programming around it), VHS
proved the more endearing, and enduring, medium. And LD? Long since replaced by digital
versatile disc, or DVD.
PACS developers have discovered it, too. For purposes of long-term archiving, DVD
somewhat successfully addresses concerns about backward compatibility and migration of
legacy data stored on less-capacious CDs. DVDs proponents like to claim that its
unpretentious origin as a consumer product guarantees standardization, competitive pricing
and easy availability.
DVD did indeed start out that way, when a consortium of film companies, media
manufacturers, and computer companies agreed in 1995 to develop a single recording
standard with which to control the mammoth (and obscenely lucrative) home video/recorded
music/computer applications markets. The media manufacturers, who up until then had been
merrily churning out incompatible digital disc formats, organized themselves into the DVD
Forum under crushing persuasion by Time/Warner, Sony, Microsoft, Intel, Apple and IBM.
And so the happy DVD Forum drafted the DVD Specification Guidebook, the collection of
technology standards by which every manufacturer abides, sort of. The fact is that DVD, at
the ripe old age of seven, is evolving at a blinding pace. Despite glowing claims to the
contrary, compatibility of its past and future incarnations is hardly a forgone
conclusion.
There are three key types: read-only, recordable and rewritable, and thats where
anything basic about it ends. DVDs detractors are quick to argue that it is not yet
a mature enough medium for enterprise use. They may have a point.
DVD-ROM is a high-capacity version of a CD. It became affordable to consumers in 1999,
and it began outselling CD-ROM fivefold once companies found they could use it to market
movies, video games, encyclopedias, phone databases, and the like. DVD-R is the
write-once/read-many version. DVD-RAM is the first rewritable version (introduced by
Hitachi, Panasonic, and Toshiba). Panasonic (a division of Matsushita Electric Corp.,
Secaucus, N.J.) developed a technology that made its DVD-RAM hugely faster others.
Unfortunately, the early versions (circa 1998) wont read double-density DVD-RAM
discs, which were introduced right afterward.
DVD-RAM has since been joined by DVD-RW (developed by Pioneer Electronics, Long Beach,
Calif.) and the rival DVD+RW (collaboratively developed by Sony, HP, Philips Electronics
and Yamaha).
DVD+RW drives will record either video or random-access content, and DVD+RW media can
be read by both DVD-video players and the DVD-ROM drives in PCs. DVD-RW is somewhat less
compatible with video and PC drives, but its cartridge-based medium makes it more suited
to archival applications than to consumer devices.
First-generation DVD+RW will not support DVD+R, although HP and Philips now sell
second-generation drives that do. DVD+RW, DVD-RW, and DVD-R drives will neither read nor
write to DVD-RAM media, although DVD-RAM drives will write to DVD-RW.
Furthermore, DVD-R comes in two types: DVD-R(A) (Authoring) and DVD-R(G) (General).
Each has its own recorder, and they are not interchangeable. DVD-R(G) records with
encryption coding that prevents the discs content from being copied; DVD-R(A) drives
and media are more expensive than General. There also are different DVD types for audio
and video.
Additionally, DVD-RAM media come in two styles: Type 1 is sealed in a non-removable
cartridge; Type 2s cartridge can be removed for playback in CD-ROM drives. Most
DVD-RAM drives are incompatible with earlier DVD-ROM drives. Fortunately, however, many
DVD players will play old CD-ROMs, CD-Is, video CDs, and sometimes CD-Rs, although DVD-ROM
drives wont always read CD-R media.
Theres more: Pioneer DVD products are famously incompatible with all others. To
refute this notion, Pioneer commissioned a study last July by Intellikey Labs (Burbank,
Calif.). Testing involved 100 domestic and international drives ranging from brand new to
five years old. The results, released in September, showed that DVD-R and DVD+R were each
78 percent compatible with all formats, while DVD-RW was 58 percent compatible, and DVD+RW
was 63 percent compatible. The test concluded what everyone suspected anyway: The newer
your drive, the better your chances are of reading something with it.
For now, though, DVD+R and DVD+RW record 20 percent faster than DVD-R and 140 percent
faster than DVD-RW, although the latter is cheaper to buy, easier to obtain, and really
holds a lot of stuff. For cardiac, its just the ticket.
The Encompass archive from Heartlab, Inc. (Westerly, R.I.) utilizes the DVD-R
write-once technology from Pioneer. Why not hard-drive or tape archiving, or even MOD?
Spinning discs are electromechanical devices. They fail all the time, says
Matt Aitkenhead, Heartlabs V.P. of technical operations. You want something
that cant be damaged easily or erased. Thats one of the problems with tape;
its rewritable. Just because a medium is fast doesnt mean its a
well-designed archive. Most medical archive technologies utilize a third-party piece of
software not specifically written for managing big cardiac images [often a gigabyte or
more]. Theyre just a generic jukebox manager. It doesnt matter how fast the
mechanism is; thats very inefficient.
Aitkenhead says that due to consumer demand for high-definition TV, which requires more
storage than todays DVDs offer, the DVD Forum is very aggressively developing
higher-capacity DVD. We anticipate the price per storage unit to be far below that of
anything else. Its really going to change the medical storage story.
In September, Sony cranked everything up a notch by introducing Dual RW, the first
drive to support both DVD+RW and DVD-RW. It also supports 24X CD-R and 10X CD-RW using
high-speed CD-RW discs. Though intended for the consumer home-video-to-disc market, it is
also the first drive to support 4X DVD-R (a much-anticipated rewritable medium that will
be on the market soon), making Dual RW the fastest DVD rewritable drive available. It will
likely find its way into enterprise systems soon.
Everyone with 8-track players, raise your hand
Of course, papers a pretty good archiving medium too. CLIR says it lasts up
to 500 years, and microfilm up to 200 years, while some magnetic tape [and] many
optical disk media
are not reliable after five years.
Its difficult to argue. If the office intern uses a paper file for a coffee
coaster, you may sustain a blotch. Park hot java on the file server, and your data are
history.
Electronic storage is more fragile than vendors want you to know. Magnetic media can
suffer from degradation of signal level from age and head wear, and magnetic fields can
easily corrupt their contents. Optical disks are susceptible to surface scratches;
breakdowns can occur between the layer where data are stored and the protective layer;
also, the protective layer can cloud up and prevent reading. Tapes tend to oxidize,
corrode, become brittle, and delaminate when their magnetic coating separates from the
backing. Cassettes will break if dropped. The National Institute of Standards and
Technology (Gaithersburg, Md.) says a 20-year lifespan is as much as you can expect for
magnetic tape, even under ideal storage conditions.
The bigger challenge to data preservation, however, is in how long any given storage
medium will be of any practical use. In Into the Future, CLIR points out that in 1976,
when the National Archives requested data from the 1960 census, it took the U.S. Census
Bureau three years to comply. Why? Because it no longer had machines that could read the
16-year-old magnetic tapes.
To get an idea of what tomorrows deadwood will be, just talk to Yves Martel,
president of TomoVision (Montreal, Quebec). He developed SliceOmatic, specialized software
used by medical researchers to analyze CT slice sets. Their problem? They have new
PAC systems that will only accept DICOM images, explains Martel, theyre
stuck with all these old magneto-optical discs or tapes or whatever, and they need to
transfer the data. Thats why Martel designed a companion program called
ReadOmatic. It deciphers a wide range of abandoned proprietary storage formats.
The one my clients ask for most is the DAT format from GE; the second would be
the Siemens format. Most of the people who have old magneto-optical discs have Pioneer;
the Pioneer drives only read Pioneer discs. Then you have the old half-inch reel tapes [a
50-year-old technology], and QIC [quarter-inch cartridge] produced for UNIX and used up
until a few years ago.
Sound familiar? Then youll be glad to know ReadOmatic is available free on
Martels Web site, www.tomovision.com.
Want to migrate data from one type of tape to another? Its usually a fairly
big deal, says StorageTeks Hayakawa. Theres a software we resell,
called ASM [Application Storage Manager], to move it using the automation of the library.
Without it you end up having to do manual media migration, which is very painful if
youre talking about 4 or 10 or 20 TB; you just dont even want to think about
that.
So where does that leave media with 3 to 10 decades of shelf life? The 30-year
retention of tape media is valid [with] proper storage and handling. Round reel tapes have
lasted for almost 30 years, says Monte Holmes, senior account manager of the data
storage group at Ovation Data Services, Inc., a Houston data recovery company.
Unfortunately, no tape drive technology or embedded format will be available for 30
years. As disk drive storage capacity requirements increase, tape capacities increase
rapidly, leaving all tape technology obsolete, often in less than a couple of years. The
chances of reading any tape media after 30 years without duplicating the original
environment are very low.
Last one on the holodeck is a rotten egg.