Does DNA Hold the Keys to the Future of Data Storage?

Last Friday, researchers from Harvard published a paper describing how they were going to change everything in IT and the computer industry with just one new technique. Rather than always relying on data stored using clusters of magnetized grains on a spinning platter, they described a way to attach (fake) human DNA to a microchip to store so much data you could stash every byte on the Internet into a single thumb drive.

Other than seeing microchip and DNA and thinking it was a cool neuroscience experiment, you may have missed the story. Because it's storage. And storage is boring, especially for people who spend their days working on networks or software or in any other geek specialty.

The breakthrough, in a nutshell, was this: A team of Harvard geneticists working on ways to create a complete, entirely artificial chain of human DNA also discovered a technique to use DNA microchips to store ridiculously high volumes of data in a ridiculously small space.

DNA microchips, also called DNA microarrays, are microchips with strands of artificial DNA embedded in them, linked to the underlying circuitry and fused in place with melted plastic.

They're used mostly in the biotech industry to figure out which genes respond to which stimuli without having to have live cells or live patients right in front of the experimenter.

What DNA does best, however, is store data.

By using DNA sequencers for things they weren't designed to do, researchers are able to record tons of regular data within DNA--customer records, for example, as well as instructions for whether your eyes should be brown or blue.

Using their own techniques--which involved an inkjet printer that produced the synthetic strands of DNA and a couple of days each to do the write and read of data onto the microarrays--the Harvard team was able to store the full 5.27M bytes of data that made up a large genetics text book onto 55,000 strands of DNA, representing less than a thousandth of a gram of material.

Top-quality hard drives can store about 25G bytes per square inch, or between 5G and 6G bits per cubic millimeter, according to a paper in the journal Solid State Technology.

The DNA microchips ginned up by the Harvard team can hold 5.5 petabits, or 5.5 million gigabits per cubic millimeter. That's roughly 110 million percent more than a good hard drive or flash drive can manage today.

Keep in mind these numbers are purely imaginary simply projections of future capacity based on reports from an experiment that succeeded in storing 600 times as much data on one microarray as anyone ever did before, and whose results may never be replicated.

Of course, engineers and storage vendors being the way they are, it's likely that if the Harvard team's results are never matched it will be because commercial R&D teams beat those results like a dead horse, not because they couldn't reach the mark.

Either way, if DNA microarray storage turns into a real, practical, stable, cost-effective way to store data, it will knock the pins out from under the single factor limiting changes in the design of digital hardware--the size of its onboard storage.

Next: The Smaller the Better, When It Comes to the Future of Data StorageRemember how cool the original iPod was in 2001? Remember how tiny it looked, considering the 5G bytes of music files you could put on it? It was big, though: 7.62 square inches total. In September 2010, the sixth-generation iPod Nano shipped, with 8G bytes or 16G bytes of storage space and a total physical size of 0.8 square inches.

Some of that size difference came from the IT industry's 60-year campaign of miniaturization, which started in a building full of ENIAC--a computer so large actual moths could fit inside to cause shorts and glitches and had to be removed by hand ... removed by technicians who could also fit inside ENIAC.

Get the comparison now? If DNA microchips do actually have a future as the highest-density digital-data storage devices ever, it won't take 60 years to turn them into commercial products. More like five years, or six.

Remember all the reports during the past few years about the rush to build new data centers because the old ones are overwhelmed? Gobs of storage are needed for the images, backups, configurations, policies and data for virtual desktops, virtual smartphones, virtual servers, virtual applications, virtual infrastructures and virtualized everything else in the consumerized enterprise.

What if all the space devoted to network-attached storage and storage area networks was suddenly not necessary because servers, NAS boxes and storage arrays had all shrunk down to one-one-hundredth of their original size? Or one one-thousandth?

Using the data density they already achieved, the Harvard team estimated in the paper they published in the Aug. 16 journal Science that they could not only fit the Internet on a flash drive, but also that they could fit all the digital data stored anywhere in the world onto chips using only 4 grams of artificial DNA, printed out using an ink-jet printer.

The size of the chips, efficiency of the storage process, its speed and the amount of synthetic DNA on each microchip will change too much by then to predict how many chips or how much space that would require.

Let's just say it would let you stick your whole SAN in a storage closet, if not on a single shelf.

Whether you'd call the resulting devices handheld, mobile, ubiquitous or anything at all when they're as invisible and omnipresent as air, none of them would require the same designs, desk layouts, physical-network wiring infrastructures or, almost certainly, the tangle of wires, pizza-boxes and chassis that pack the average server closet.

Next: IT After DNA Storage: Think SmallInformation technology would become even more ubiquitous but, also, almost invisible. It would use a fraction of the electricity as a typical PC or laptop and generate so little heat that data center exhaust would no longer be sufficient to heat a small village or cook a turkey in just hours.

The devices would save money and store data more reliably (except for the inevitable DNA mutations and resulting superheroes made from live data).

If they become practical, DNA microchips--or, less likely, holographic or other optics-based storage media--will do more than just shrink our hard drives. They'll change the way our computers look, where and how we take them with us, how we use them, how we structure the physical spaces of our workplaces, the IT infrastructures that feed data to us and the budgets, headcounts, specialties and roles of the IT people making it all work.

DNA-based data storage is unquestionably one of those stories you see and assume you'll see it about the time you put your first real jetpack or flying car in the driveway. It's the kind of story you read and think "Cool," then forget about.

I don't think that will be the end of it, though.

The endless march of miniaturization has us all programmed to expect smaller, more powerful computers in more innovative shapes every couple of years. Every once in a while there's a big development--like virtualization--that produces a serious change in the number of physical boxes that pack corporate data centers and the to-do lists of the IT people who work there.

It's an attractive and oft-repeated thought that storage, like all components of computers, can be miniaturized to the point that "the computer" largely disappears, except for the accessories we use to make the data bits dance.

That's more than just a bit of futuristic nonsense, though. It's a concrete expectation, and, with this report on the incredible density of storage in DNA microchips, it's one that we could start planning for already.

Think virtualization and cloud changed the way your IT shop is physically laid out? The way the staff is organized? The number of times you have to go touch a machine to make it work right? That's nothing. It's a change of paint, a rearrangement of the deck chairs in the sauna-like atmosphere of the data center.

Shrink your storage down far enough, and you're no longer limited by the size of your hard drive or the capacity of your solid-state drives. You're limited only by your imagination and the need to do something useful with the square miles of heavily climate-controlled, highly secured, empty spaces in what used to be your data centers.