Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ High ASPirations ]   |  What's in Store  |   [ Invincible ] >>

Weekly Column

What's in Store: How IBM's Departure From the Disk Drive Business is Likely to Change an Entire Industry

Status: [CLOSED]
By Robert X. Cringely
bob@cringely.com

Maybe you saw news stories this week about a study from the University of California at Berkeley predicting that the world's data storage requirements will double in the next two and a half years. Okay, that makes sense, except the stories tended to put it more like, "Human knowledge will double in two years!" This is, of course, nonsense. What will double is all the garbage we store on our computers, which is to say not much of real value. I shudder to think how much of this growing pile of stored bytes can be attributed to bootleg Britney Spears songs alone. But at least it got me thinking about data storage — a topic hardly anyone thinks about these days.

Part of the reason why our data storage requirements are increasing so rapidly is because more and more data is in digital — rather than analog — form. The Berkeley study, for example, has to take into account all the music CDs in the world. In the days of vinyl records, we wouldn't have thought to include private music collections as part of the world's data storage. But those days wouldn't have seen a company like EMC sponsoring the UC Berkeley study, either. Storage is big business.

Storage is divided into big systems (like those from EMC aimed at large corporations and government) and small systems (like the new hard drive you just bought at CompUSA). Storage is also divided into online and offline. Online is available to you immediately on your PC or over the Internet, while offline requires mounting a tape or finding a floppy or CD.

Though few people realize it, and the storage industry prefers to ignore it, the real point of vulnerability for data storage is offline. Most archives and backups are held on magnetic tapes, and tape drives haven't kept up with the progress of magnetic and optical disks. While areal density — the number of data bits that can be stored per square inch of storage media — has been steadily increasing for magnetic and optical media, it hasn't changed at all for tape drives. Instead, to make tapes hold more we just make them longer. And to allow those longer tapes to fit in the same size cassettes, we have to make the tape thinner. The result is more data density, but it comes at the cost of tapes that are more fragile and often can't be reliably read by any drive other than the one upon which they were originally recorded. This is a point of vulnerability. You have data backups, I'm sure, but are they even readable? Many are not.

In a world of web surfing, it is interesting to consider the efficacy of offline storage at all. How long are people willing to wait for data that isn't already online? IBM did a study and the answer, sadly, is three seconds. If you want data, but it requires mounting a tape or searching a network, that tape had better be up or the search result retrieved in three seconds or less. Otherwise people simply lose interest. We are so shallow.

I think this argues for keeping all data online at all times.

Any look at the future of data storage brings us to consider wacky technologies like holographic storage and other schemes that will increase storage density by the several orders of magnitude needed to meet our gluttonous long-term storage needs. These new technologies will come in time, but for the moment, magnetic storage is king for immediate access, and optical is tops for long term storage, and that isn't likely to change in soon. Right now, magnetic areal densities are approaching 50 gigabits-per-square inch in shipping products and 100 gigabits in the lab. The probable maximum density for magnetic storage using current technology is 500 gigabits-per-square inch, which Moore's Law says is good for just about another five years. So don't look for a dramatic change in storage technology until 2007.

One area where the storage industry is particularly vulnerable is research and development. The tendency has been to let IBM do the research for the whole industry. The GMR recording heads that are used in all of today's big disk drives are a good example of this. IBM invented GMR heads, but rather than keep the technology to itself, Big Blue licensed its GMR technology to all comers. IBM simply preferred to make its money from licenses, and the other drive makers were perfectly willing to let IBM assume the risks of research. But now IBM has sold its PC disk drive business to Hitachi, and there is no way of knowing whether the Japanese company will be so generous with its technology. They probably won't be generous, preferring — in typical Japanese business fashion — market share to license revenue.

This means the companies that previously depended on IBM technology will have to find their technology elsewhere or invent it themselves. Either route is more expensive in the short run, and fraught with risk if the big competitor's research program works out and yours doesn't. This means that companies like Maxtor and Seagate will have higher costs and somewhat greater risk, which will lead to both greater stock volatility and probable downward pressure on profit margins that are already painfully low.

The likely outcome, then, of IBM's decision to exit the small drive business is a major consolidation of all IBM's former competitors and licensees. Maxtor and Seagate will survive, but smaller companies like Western Digital will have to grow or die. And growth, in this case, means vertical integration. WD is going to have to acquire the bits it needs to reach the point where it can develop all its own technology in-house. So look for disk drive technology companies like Read-Rite to be sucked up.

This doesn't sound very rosy, does it?

If I was writing this column two years ago, I might be covering the whole area of Internet data storage, for that was at a time when you could get free data storage on web sites — a business that has pretty much gone away. And what's interesting is not that those businesses existed, but what lessons they taught us through their departure. If you hand over your data to some third party, what happens when they go out of business or change their policies such that the erotic e-mail you were saving to savor in your golden years is suddenly gone? Even mighty Microsoft has been whittling down the storage for its Hotmail users. They recently eliminated long term storage of read mail for certain users, making the mistake of implementing the change — erasing the messages — before announcing that they were going to do so.

Can you trust a third party to reliably hold your data? I don't, and I don't think you should, either. And this leads us, thankfully, to an area where I think there is considerable room for growth — personal private and mobile data storage. Why shouldn't we be able to take with us all the data we really value?I think the data we really value is a small subset of the data stored in our PCs. I was amazed to learn, for example, that user data typically comprises less than 10 percent of the data on most PC hard drives. The rest is application and system code. Think about it. You have a dozen or more applications and probably the installer for each on your system, yet how many letters or reports have you actually done with that word processor, and what are the total storage requirements for all those spreadsheets for your failed Internet startups? It is a lot less than you think.

I have on my main system every word I have written since 1992, which is around three million words. I also have every e-mail worth keeping, a couple databases, and many spreadsheets and Powerpoint presentations. Uncompressed, it adds up to less than 200 megabytes. Heck, that is small enough to fit on one of those USB flash drives that attaches to your key ring!

As an opinion leader, then, it is my intention to carry with me all my data all the time. Let the house burn down, I'll still have a career. For knowledge workers, data storage means everything, but it doesn't have to mean carrying with you an 80 gig FireWire drive.

So I see a very interesting niche developing for personal data storage devices, especially those that are password protected. Two products in this line I'll commend to you are Storcard and JMTek's USBdrive. Storcard, backed by the legendary Finis Connor, puts up to five gigabytes of encrypted data on a piece of plastic the size of a credit card. JMTek's USBdrive puts e-mail client software on the typical key fob flash drive so that you can plug your USBdrive into virtually any Windows computer and have your native e-mail application up and running in seconds, complete with all your stored mail.

Comments from the Tribe

Status: [CLOSED] read all comments (0)