Skip to Content Java Solaris Communities Partners My Sun Sun Store United States Worldwide

»  Contrarian Minds Archive

The Trouble with Tape

Hint: It's nothing to do with the technology itself.

By Al Riske

08.Jun.06--People have been telling Richard Dee that tape storage is dead since 1984.

Optical disks are the wave of the future, they'd say. Ignore the trend and you'll be out of a job before you know it.

"Well, it's now 2006 and tape is doing just fine," says the Sun Fellow.

A physicist by training, Dee joined Sun when the company acquired StorageTek, a leading provider of archival storage (read: tape), last year.

"The whole of society likes trendy things and tape isn't trendy, but it does the job," he says. "That's why it's still here today and why it will still be here 20 years from now."

"Back in 2001, we said, 'Oh, we can put a terabyte on a cartridge no problem,' and nobody believed us. Well, you know, what? It's coming out next year."

Richard Dee
Sun Fellow
Sun Microsystems

 

The only trouble with tape, he says, is a slight image problem. The technology is thought of as old and slow.

Well, it is and it isn't.


Tape storage has been around for 50 years, so, by high-tech standards, it's ancient.

Look closer.

"The stuff we're using today, and will be using in the future -- there's not anything old about it," Dee says.

What people are overlooking, he suggests, is the not-so-subtle difference between a Model T Ford and the latest BMW. Both are cars. Both use internal combustion engines. But a lot has changed over the years.

Consider tape capacity.

"In 1986 the capacity of a tape was 200 megabytes. In 1996 it was 5 gigabytes. In 2006 it's 500 gigabytes," Dee points out.

"Back in 2001, we said, 'Oh, we can put a terabyte on a cartridge no problem,' and nobody believed us. Well, you know, what? It's coming out next year," he adds.

"The next thing is, we'll say, 'Okay, we can put 10 terabytes on a tape.' Does anybody believe that? Probably not. It turns out I think we can."

"Imagine copying 100 photographs from one device to another in a second. That's the speed these tape drives run at."

Richard Dee
Sun Fellow
Sun Microsystems

 

Dee is perhaps best known for developing innovative tape heads that enabled StorageTek products to break away from the pack.

"StorageTek didn't do their own tape heads initially; they used to buy them. But the technology had to move on and the people who were making the heads at the time didn't move on with it, so I was brought in," Dee recalls.

He was the lead engineer on most head designs during the 1980s and '90s, before moving into R&D in 2001.

"Our latest product has 32 parallel channels, each running at four megabytes per second. All together, that's 128 megabytes per second. Imagine copying 100 photographs from one device to another in a second. That's the speed these tape drives run at. It's because of the multichannel magnetic structures we have the ability to design and build here," Dee says.

"The next thing is, we'll say, 'Okay, we can put 10 terabytes on a tape.' Does anybody believe that? Probably not. It turns out I think we can."

Richard Dee
Sun Fellow
Sun Microsystems

 

The really big changes in data storage were made possible, in part, by the Internet.

"When people wanted to interchange data they started to send it through the net instead of sending a tape. That changed the game because we didn't have to be compatible with the other guys; we could do something better," Dee says.


He notes, however, that tape is still used as an interchange medium -- at a much higher level.

Which brings him to the set up for what he calls the standard tape joke.

"Tape is seen as the same old, slow, lumbering technology. You can't get to data quickly. That's its big downfall. Because it's a serial access device, if you want a particular file, you've got to wait a minute rather than a millisecond. However, if you want to send a petabyte of data from here to Australia, it's quicker to put it on tape and ship it by boat than it is to send it by gigabit Ethernet," Dee chuckles.

"When 10-gigabit Ethernet comes out, the joke will be, you put your box of tapes on a 747 rather than a boat, but right now a boat is still faster than gig-E."

Within five years or so, he predicts, a petabyte of data on tape will fit in a suitcase that you can take with you as checked baggage.

"If you want to send a petabyte of data from here to Australia, it's quicker to put it on tape and ship it by boat than it is to send it by gigabit Ethernet."

Richard Dee
Sun Fellow
Sun Microsystems

 

Where tape enjoys its biggest advantage is in cost.

"Tape is a factor of 10 cheaper than disk, and there are people out there who buy storage by the petabyte. If you're buying disk drives, you're talking big money. Multiple millions of dollars. If you buy tape and it's a tenth of that, well, that gets your finance guy's attention," Dee says.

And that's not all.

"If you are going to store the data and not look at it very often -- you have to keep it because of regulatory requirements -- you don't keep it on spinning disks consuming power. You put it on tape and put it on the shelf. Then if you need it you go get it," he adds.

With energy costs rising, that consideration is gaining currency.

Though he concedes that tape is hardly perfect, Dee believes it's still the best solution there is. "A tape cartridge will soon hold a terabyte of data and cost you $100. You want another terabyte tomorrow, you put another tape on the shelf. That's another $100. It's a simple solution and it's scalable," he says. "You want to do that with disk, you put in a RAID system. You want another terabyte, you've got to put in another system and integrate it into your network."

"In terms of any sort of physics barriers, tape has plenty of life left."

Richard Dee
Sun Fellow
Sun Microsystems

 

Dee is clearly proud of the advances in Sun StorageTek systems.

"The technology we have in this stuff is right on the edge. If you want to get the capacities and the densities and so forth, it has to be," he says.

"For instance, the recording bit density. Even though it's not as high as disk, the bit length is a fraction of the wavelength of light. The wavelength of green light is 500 nanometers. The bits on tape are down around the 100 nanometer range now. The fact that we can actually resolve that with the recording heads we make and produce digital data storage using that on a tape is, I think, pretty good."

In fact, the advances in tape technology are pretty amazing when you consider the constraints Dee and his colleagues work under.

For one, customers don't want to change their tape format every six months.

"If you've got 100,000 tapes, do you want to go and buy them all every six months and rewrite your data to them? No. That forces the tape industry to move slowly," Dee says. "Tapes are expected to work in every tape drive ever made, within that product family. They must be able to read and write with the same error rate on millions of tapes. People say, 'Tapes have always done that. That's boring.' Well, you know what? Disks don't have to do it."

Without that limitation, disk storage has been able to advance at a faster rate (with tape following its lead), but Dee sees an ironic twist developing.

"They're running out of physics now. They've been struggling with that and tape hasn't," he says. "In terms of any sort of physics barriers, tape has plenty of life left."

Tape is dead, you say?

Long live tape.


Readers Survey
I found this article...
Not Informative   Informative   Very Informative
Comments:

Richard Dee

Title: Sun Fellow

Job: Lead technical engineer/physicist for tape technology at Sun.

Expertise: The physics of magnetic recording.

Quote: "People have been telling me tape is dead since 1984. Well, it's now 2006 and tape is doing just fine."

Claim to Fame: Design and development of innovative thin-film magnetic recording heads that differentiated StorageTek systems from the competition.

Education: Bachelor's, master's, and doctoral degrees in physics from the University of Lancaster in England.

Background: Did research at the University of California at Los Angeles (UCLA) and at the University of British Columbia in Vancouver, B.C. Worked on superconducting magnetometers using SQUIDs (Superconducting Quantum Interference Devices). Designed products and did advanced research at StorageTek from 1982 to 2005. Joined Sun when it acquired StorageTek.

Patents: 20 issued, 11 pending.

Writings: More than 50 published technical papers.

Activities: Golf. Home repair. Cooking. ("Best thing to put in sauces and gravies is Martini & Rossi dry vermouth. Not a lot, just a glug.")

Favorite Food: Rack of Lamb.

Last Book Read: The Modigliani Scandal, by Ken Follet.

Little-Known Fact: Received graduate teaching diploma, master's degree, and Ph.D. in physics, all in four years.

Pet Peeve: People who run red lights. ("It's a miracle we don't have more pile ups.")

Childhood Ambition: Wanted to drive trains.

First Job: Helping out in a dry cleaner's shop, fetching clothes off the rack and sweeping the floor, on Saturday afternoons.

Perfect Day: "I think it's being with the right people. Family and friends. That can take a bad day and make it perfect, you know."

 
Would you recommend this Sun site to a friend or colleague?
Contact About Sun News Employment Privacy Terms of Use Trademarks Copyright 1994-2008 Sun Microsystems, Inc.