Software

Shameless Self-Promotion

For nine years and 492 columns I have described new companies and new technologies always with the qualification that I own no shares and have no financial stake in them. This statement is for two reasons: 1) I am poor and can't afford to jump into gee-whiz technology deals, and; 2) my contract with PBS requires that I make it very clear if I have a financial stake in the product or company I am writing about. I am not prohibited from writing about my investments, but I have to make it clear when I am financially involved.

Well this time -- for the first time ever -- I DO have a financial interest in the technology I am writing about. In fact, the whole thing is pretty much my idea. And while my writing about it may appear to be shameless self-promotion (heck, it IS shameless self-promotion) at least I can be sure it is ACCURATE shameless self-promotion. And since I made a presentation on the underlying technology last weekend at an ACM conference, I figure the cat is out of the bag anyway.

As many readers know, I have a consulting business that predates this column by six years, and I have worked on various technology projects over that time, so this is not new behavior. It is just the first time one of those projects is newsworthy. I wish they all were, but sadly not.

This story starts a year ago when I wrote a column on the exploding electric bills of large data centers. You'll find that column among this week's links. In subsequent weeks I heard from many frustrated managers of huge hosting facilities and even had a few meetings trying to help them figure out ways to save energy and to save ON energy, which is a completely different thing. What became clear to me in those meetings is that the culprit here is storage -- the energy consumption of modern disk drives.

While processor companies like Intel and AMD had spent billions trying to lower the energy consumption of their chips, the disk drive companies had done very little to save energy on storage, other than to point to increases in areal density (the amount of data that can be packed onto a storage medium) as proof that both cost and energy consumption were dropping in terms of dollars- or watts-per-gigabyte, if not in terms of dollars-per-month. Disks can be spun down, too, but that's no thanks to the disk drive companies, either.

Since disk storage is responsible for consuming 30 percent of the energy in your PC and up to 50 percent of the energy in a data center, it seemed to me that this would be a good area to look for improvements. Add to this the problem that in this world of YouTube and Sarbanes-Oxley no data center has enough storage, it becomes clear that we need disk drives that hold more, cost less, and use less energy.

So I found one.

Two old friends of mine, Anil Nigam and Jim White, and their company, Antek Peripherals, Inc., had been working for years on technology for a sort of hyper-floppy drive using metal foil for the recording medium. At the time they were aiming their work toward digital cameras, but I asked if the same technology could be used in a non-removable form inside a computer disk drive? And if it could be used that way, what would be the effect on price, performance, reliability, and energy consumption? The answers were stunning: they could design new families of disk drives that held up to three times as much data in the same space, were more reliable, actually cheaper to build, and used 70-95 percent less energy to run than the current state of the art.

For a data center operator, simply swapping disk drives could increase storage by three times while decreasing total energy consumption by a third. This is a huge attraction for big businesses and big hosting companies, not just because it lowers their electric bills but because it may mean they can avoid building more data centers entirely.

But was it too good to be true?

Anil and Jim are no yokels when it comes to disk drives, having about 50 years of experience between them. Anil co-founded SyQuest, an early competitor to Iomega. And Jim is an academic who formerly worked on disk drive design at IBM and presently earns royalties on most of the disk drive heads manufactured in the world. And the metal foil disk technology they developed had been patented years before. Issued patents are a rarity for leading-edge technologies, but this technology had been leading by a decade, only nobody knew it.

Now we had to start talking to disk drive companies, recording head companies, manufacturers of metal foil, and of course big customers, validating the technology and lining up potential partners, which is what has occupied part of my time over the last several months. In doing so we came up with a few more technical advantages, too.

The technology in question replaces the aluminum or glass platter in your hard disk drive with a "platter" made from stainless steel or titanium foil that is 22 microns or 25 microns thick, respectively. The materials cost more but we use so much less of it (the disk is so incredibly thin) that the total material cost is substantially less. This "floppy" material has the same kind of magnetic coatings used on standard disk drives and our drives live on the same technology growth curve as those others. The way we obtain greater storage density is simply by putting more platters in a drive (say 12-15 instead of 4-5 in an enterprise 3.5-inch drive) because they are much thinner and can be stacked closer together. The only parts of the drive that are significantly different are the platters and the heads and the heads vary only in having an extra slot. There is no rocket science here, but what science there is is patented.

The advantage of our drives goes beyond enterprise applications. We are able to build cheaper drives, for example, because our platters cost less to make and the nature of our flying heads is such that dust is sucked away from the head-disk interface, meaning the drives do not have to be assembled in a clean room. Also, where current platters are individually polished then sputtered, our metal foil can be polished and sputtered in giant rolls then die cut as needed, keeping costs down and manufacturing flexibility up.

Because of the dramatically smaller rotating mass, our drives can use smaller spindle motors that cost less, weigh less, and use less energy. Our 3.5-inch drives can use the spindle motor from a 1-inch drive. For super-high-end applications we can make 30,000-RPM drives that will appeal to the Oracle and DB2 crowd. Our PC drives will cost less and use one quarter the power.

This power savings is key and opens up whole new product categories, like tape replacement. Computer tape drives are a $10 billion business, but the dirty little secret is nobody really knows if they have the data or not, since tapes are not a reliable archival solution due to print-through and environmental deterioration. Disk drives would be better but disk drives cost too much. Not anymore. Our projected tape drive alternative costs $0.20 per gigabyte to tape's $0.18, but ours has greater reliability and a seek rate that is 720 times faster. The tape drive uses 18 watts while our tape replacement drive uses three watts.

The nature of our drives is such that they are very resistant -- almost immune -- to shock damage, making head crashes a non-event because the flexible metal foil yields to the head, pushed away by a layer of compressed air, rather than being struck by it. So our drives are perfect for notebook computers and portable music and video players because they are lighter, tougher, and make batteries last 70+ percent longer.

In an iPod, for example, our 60-gig drive would be the same size as the iPod’s 30-gig drive, but ours wouldn't need head-parking or "uh-oh I'm falling" circuitry, so it would be cheaper to build. And while that 30-gig drive takes five seconds to spin up for each gulp of music, our 60-gig drive spins up in 0.4 seconds. Map the area under that power consumption curve and you'll see that battery life can be extended dramatically or smaller and cheaper batteries can be substituted.

But ours isn't the only new disk technology coming to market. Windows Vista will support the new hybrid disk drives that add flash memory as a sort of L4 cache that can be read first, either allowing the platter not to spin up at all in certain circumstances or at least to provide faster access as the platter is accelerating. The only problem with hybrid disk drives is they cost more -- a LOT more -- just like Vista costs more than XP.

Our metal foil drive costs less, not more, and spins up so quickly that data can be read from disk as fast or faster than it can be read from flash. Who needs a hybrid disk drive?

Who needs flash in general as a mass storage technology? Our 10-gigabyte 0.85-inch drive can spin up, read or write data, then shut down again, all in less time than it takes to perform the same task using flash while being just as resistant to shock damage and more resistant to heat. That 10-gig drive will cost $24 compared to $240 for 10 gigs of flash, so we expect that our technology will be used for any application requiring more than 2-gigs of storage. The obvious market here is mobile phones, which will become media storage devices.

The market potential is one billion computer disk drives and one billion mobile phone drives per year. And it all starts around this time next year when metal foil drives will begin to appear under well-known brand names. You see I'm not really starting a disk drive company. Now THAT would be stupid.

A Tradition of Empty Boxes

Congratulatory e-mails started arriving on Monday as more than 100 readers eventually told me about Sun Microsystems' Project Blackbox, which is a data center in a shipping container. Announced officially on Tuesday, Blackbox is essentially the Google shipping container data center I described nearly a year ago, finally realized as a product. It's nice to be correct, of course, but are these two products actually one in the same? I think they are.

Google doesn't actually need Sun to build its mobile data centers, of course, but on the other hand, why not? Someone has to build them, and this new form factor is such that it will require a real assembly line, which Google may not want to build. But more likely this is Google Chairman and CEO Eric Schmidt throwing a bone to his old company, where he was the longtime CTO before going first to Novell, then to Google.

There is a long history of customers building their own computers and, in fact, that's how Sun got its start as you can read starting in the next paragraph. Why Google isn't especially bothered by the idea of sharing this new form factor with present and future competitors is because the hardware and the shipping container don't really matter all that much. It's the software and data that matter to Google, and Sun has no right to license those. Ironically, this is a 180-degree flip from the attitude toward intellectual property that existed when Sun, itself, was formed.

I love the story about how Sun was founded as told to me long ago by Ralph Gorin, who for many years ran the computer networks at Stanford University. According to Ralph, Xerox had installed a couple Alto graphical workstations in the Carter White House, where they were noticed by resident spies from the Central Intelligence Agency and National Security Agency. These spooks wanted GUI workstations, so they sent their IT people to Palo Alto to buy some from Xerox PARC.

Nobody really knew what an Alto cost, since the workstation had not been developed to sell. You could price out the components, of course, but that would pay nothing for the system software. So someone at PARC, sensing a lucrative sale to a customer known for buying $900 hammers, $1,300 toilet seats, and occasionally launching private wars on banana republics, decided to hit the feds with a huge bill -- one so staggering that not even the spies would consider paying it. Before flying back to Washington, however, they drove the two miles to Stanford for lunch with Gorin.

There they saw the Stanford University Network (S.U.N.) Workstation as designed by grad student Andy Bechtolsheim. The S.U.N. Workstation was a poor man's Alto with a robust version of Unix, a graphical user interface, local storage, mouse, and Ethernet. And the price was right, too -- that is if Andy B could get anyone to actually build the machines that the feds wanted to order.

Stanford, itself, wanted no part of the deal. Gorin didn't see the university as being in the computer business. University lawyers looked inside the S.U.N. Workstation box and decided that IT CONTAINED NO INTELLECTUAL PROPERTY. So they gave Andy permission to sell the boxes if they were built by an outside vendor using no university resources.

Notice the complete about-face here. Today, Google sees nothing strategic in the Blackbox hardware while 20+ years ago Stanford saw nothing strategic in the S.U.N. Workstation hardware OR software.

Poor Andy never found a manufacturer, either, though he visited every company of note in Silicon Valley including 3Com and IBM. The IBM meeting was especially important, so Andy, who did not own a suit, went for help to the Stanford Drama Department where they fitted him with a set of white tails for his "formal" meeting with IBM in San Jose. But since the Drama Department didn't have any size 13 formal shoes, Andy wore his sneakers.

For some reason, IBM didn't go for the manufacturing contract so Andy had no choice but to found his own company to build the boxes and thus Sun Microsystems was founded.

More than 20 years later Andy is back at Sun building leading-edge servers that may well populate the Blackbox. And it ought to be a lucrative market -- one of those unusual markets that didn't really exist until someone (in this case the guys at the Internet Archive, not Sun or Google) thought of it. Few people saw the need for such a mobile data center until one actually existed and now Sun will sell a ton of them -- in almost every case without hurting sales of its existing hardware, which means this new gizmo is good for business.

The beauty of a shipping container data center isn't just that it operates stand-alone and can be plunked down in the parking lot of your existing data center or dropped by helicopter on the roof of your headquarters building. A great proportion of its beauty lies in the shipping container's efficiency not as a server but as a network. It's the largest sneakernet ever built. Moving a petabyte of data across the country using even the biggest optical fiber connection could take weeks, but the Blackbox can be installed in at most a few days.

Companies with huge data centers will use Blackboxes like school districts these days use portable classrooms, distributing them as the computing load requires with installations that will be called temporary but may well end up being permanent, at least in terms of computer-years. And the part Sun really hopes for, of course, is that big customers will keep a Blackbox or two around just in case of emergencies. At $2 million per container, a couple hundred standby units mean real money to Sun, which could use it.

Expect, too, to see a new business appear with companies renting Blackboxes. And of course we'll see other vendors than just Sun building these things. After all, if you look inside there really is no intellectual property in there, so Sun can hardly claim an exclusive.

But one mistake I think Google and Sun have made is in the form factor, itself. A smart competitor to Sun (my guess is Google) would get to work on a Blackbox equivalent that doesn't fit in a shipping container, but rather in an airfreight container or Unit Load Device. The second largest of these puppies, the LD8, holds 243 cubic feet, which is about one quarter the volume of Sun's Blackbox. Still, you could stuff an LD8 easily with a thousand Opterons and half a petabyte of storage and instead of delivering it by sea at 10 mph or by truck at 60 mph, you could deliver it by air at 600 mph.

Not surprisingly, Page and Brin's Boeing 767-200 GoogleJet can carry seven LD8s, making it the fastest networking device ever built.

Figures.

The Old New Thing

Everyone knows by now that Google is buying YouTube for $1.65 billion in stock and most everyone sees the purchase as a risk for Google. Sure, YouTube has 100 million video downloads per day (as the site constantly tells us, though PaidContent seems to think it is more like 40 million downloads per day) but a decided lack of video advertising and harping over copyright infringement by various movie studios and TV networks suggests to most pundits that YouTube will be a drain on Google for sometime to come.

Apparently not.

YouTube was fooling us. The company allowed we pundits to think it was broke because it worked to YouTube's advantage to be perceived as broke. It was newsworthy, for one thing, and YouTube loves being in the news. Being broke made YouTube appear to be an underdog to the bigger boys like Google Video and Yahoo Video and consumers often like an underdog. And being broke strongly implied the company had shallow pockets, making it a less-than-desirable target for copyright infringement charges.

Only it turned out not to be true at all. YouTube runs at a considerable profit.

I can't take the credit for figuring this one out, though I sure feel foolish for having not done so. The big brain on this case is blogger named Froosh writing at HipMojo.com (it's in this week's links). Froosh throws out some wild-ass numbers that suggest YouTube has gross revenues of around $7.5 million per month against expenses of less than $2 million.

Frankly, I think the real revenue numbers are lower and the expense numbers are higher, but even if YouTube is just breaking even, it is doing vastly better than most of us thought.

Froosh's numbers are too high and were trashed by ZDNet based on that outfit's evident ignorance of what video streaming actually costs to do. ZDNet seems to think it costs about $0.85 per gigabyte for YouTube to download video through Limelight Networks. Frankly, I don't know what Limelight charges YouTube, but I know DataPipe charges NerdTV only $0.20 per gigabyte and I am not a great negotiator, nor am I using the kind of bandwidth YouTube is. Based on my NerdTV experience and YouTube's enormous volume, I would have guessed YouTube's costs at more like $0.10 per gigabyte.

The real news here isn't that YouTube is profitable and probably banking a couple million bucks per month, it is HOW YouTube is profitable, which is through banner and contextual ads. While we were all looking for pre-roll and post-roll video ads, and lamenting their absence, YouTube was making its money the old fashioned way just like nearly any other web site.

Just like Google, in fact, which is why the fit between the two companies is good and the price, well the price doesn't matter, since it is only Google shares, which have never been held to any realistic standard before, so why should they be now?

YouTube is less a video-sharing site than it is a social networking site based around video. The company gets its content for free and has built a profitable business from acting as a video portal. In the strictest sense this isn't Web 2.0 at all and actually harkens back to Web 1.0 applications like the Motley Fool.

So is YouTube the future of television or isn't it?

The answer to this question has to be based more on how WE use the medium rather than on how it uses us. I don't think YouTube as it exists today is even remotely the future of television, because if it is then television is in huge trouble.

Where are the folks who watch YouTube six hours per day? They exist, I'm sure, but there was a time when Americans watched television an AVERAGE of six hours per day, which is more than YouTube will ever know. YouTube gets stale for me after about 20 minutes, which hardly makes it the next Seinfeld.

If YouTube has 100 million video downloads per day and if those downloaded videos average 2.5 minutes in length as PaidContent suggests, then the average daily YouTube audience consumes 250 million minutes or less than ONE MINUTE OF VIDEO PER U.S. RESIDENT PER DAY. That's a lifestyle change, true, but not a big one, nor is it even guaranteed to be permanent.

The whole television viewer experience has always been based on two factors: immediacy and production values. TV brought us live events we could share as a nation. YouTube can't do that. TV brought us production values beyond what we could afford as individuals. YouTube doesn't do that unless it is by ripping off copyrighted content.

Its evident profitability is how YouTube is now able to cut revenue-sharing deals with record companies and TV networks. There has to be revenue to share for revenue sharing to work. But if you look closely at those deals, they also involve the prospect of original YouTube-only content from partners like CBS.

Is CBS going to put $1 million per hour into its YouTube content? I don't think so. CBS is going to throw on YouTube all its old pilots and episodes it had previously written off -- content that costs it absolutely nothing because it was paid for long ago out of a different budget.

YouTube is the factory outlet of commercial television, at least for now.

The fact that YouTube isn't the future of television doesn't mean that television has no future. We are constantly finding new ways to view old content and will eventually find new content to view in those new ways, too.

I think it is ironic that YouTube doesn't carry full NerdTV episodes because the shows are too long and too big to meet YouTube specifications (max 10 minutes and 100 megabytes). Longer form videos like NerdTV that serve niche audiences are more likely the future of television than what is presently on YouTube, though of course that could change over time.

Right now the numbers say video is more of a gimmick and the substance of YouTube is actually its social networking. It isn't the new new thing at all, but more of the old new thing.

And speaking of new things, in two weeks this column will have a completely new look. This will be our fourth redesign in nine years, but this time we're changing almost everything, even the underlying code base, which is going from Jurassic HTML to a Movable Type blogging platform.

Yes, I, Cringely is becoming a blog. I fought this for a long time but the economics of change are evident here and many readers will appreciate the new commenting capability you will all shortly have. But just as many of you will hate the whole darned thing, not because it is bad but because readers nearly always hate change.

I say "Bring it on."

Risky Business

Last Saturday the United States Congress passed a port security bill that carried an amendment banning Internet gambling. This was a huge mistake, not because Internet gambling is a good thing (it was already illegal, in fact), but because the new law is either unenforceable or -- if it can be enforced -- will tear away the last shreds of financial privacy enjoyed by U.S. citizens. The stocks of Internet gambling companies, primarily traded in the UK, went into free-fall as their largest market was effectively taken away. I don't own any of those shares, but I guarantee you they will fully recover, which is part of what makes this situation so pathetically stupid.

Ironically, many of the senators who voted for this legislation may not have even known the gambling bill was attached, since it didn't appear in the officially published version of the port bill. But such ignorance is common in Congress, along with a smug confidence that people and institutions can be compelled to comply with laws, no matter how complex and arcane. The amendment was a surprise late addition, pushed by Senate Majority Leader Bill Frist, who has presidential ambitions and reportedly sees this battle against Internet gambling as part of his eventual campaign platform.

Only the new law isn't really against Internet gambling at all, since it specifically authorizes intrastate Internet gambling, imposing on the net the artificial constraint of state boundaries. So the law that is supposed to end Internet gambling for good will actually make the practice more common, though evidently out of the hands of foreigners, which in this case includes not just operators from the UK but, if you live in South Carolina as I do, it also includes people from Florida and New York. Let a million local poker hands be dealt.

What the new law actually tries to control is the payment of gambling debts through the U.S. banking system, making such practices illegal (except, of course, for intrastate gambling, which probably means your state lottery). Once President Bush signs the bill, your bank and credit card companies will have 270 days to come up with a way to prohibit you from using your own money to pay for gambling debts or -- though far less likely-- to keep you from receiving your gambling profits. The law covers not just credit card payments but also checks and electronic funds transfers.

The most optimistic view of this law from the U.S. banking industry says that controlling payment by checks and electronic transfers is simply impossible and won't be enforced. Only credit card payments are seen by the banks as being practical to limit. But what if Congress doesn't want to take "no" for an answer? What if they are serious? Then the banks will have to put systems in place to examine every payment transaction, no matter how small, and determine if it is gambling related. And because there will inevitably be attempts to get around the law, such examination would go beyond simply identifying the payee to following the money further upstream and downstream and examining it in the total context of your financial activity: Is there a suspicious trend in these payments, which appear to follow every NFL football game, for example?

If you bother to read U.S. currency, the notes say they are good for paying "all debts, public and private," which is why Tony Soprano and the Cali cocaine cartel liked $100 bills so much. Ironically, if the banks are effective in controlling other gambling payment schemes, it may all come back to paper money, which is almost impossible to trace.

Is the end here really worth the effort? The United States already has strict, even draconian, controls over fund transfers that might potentially be used to pay for terrorist activity. Buy a house or open a brokerage account and see how deep an interest the bank takes in where the heck your money is coming from. Now it is proposed that they apply the same diligence to transactions as small as one dollar.

This is ridiculous, not just because it is an unwarranted invasion of privacy, not just because we as consumers will ultimately have to pay for the cost of snitching on ourselves, but because the system of regulation ultimately won't work. With an Internet gambling market approaching $20 billion per year, there is a huge incentive for new enterprises to spring into being specifically to get around this law. Frankly, it ought to be easy.

Just off the top of my head I can think of several possible approaches to subverting this new law. Working within the banking system it might be possible to aggregate payments to make their individual origins less obvious, especially if the aggregation involves some non-gambling money. Remember, these restrictions are being placed on the U.S. banks, not their foreign counterparts, so any bank in the Caymans or on the Isle of Man ought to be able to chug through such aggregated payments without violating any local laws. Another option, since intrastate gambling is authorized, is to make interstate and international gambling debts effectively local by creating thousands of local virtual bookies. All of these are old school ideas that don't even need technology to implement. What if we bring to bear the capabilities of Web 2.0 and create payment mashups by the dozen – little PayPals that rise and set like the Sun?

Any random group of 535 nerds is smarter than the 535 members of the U.S. Congress and able to circumvent ANY regulation if there is enough profit incentive to do so. Well the U.S. Congress has just created such an incentive where there was none before. And once these various payment schemes start appearing, what's to say some of them can't be equally used to finance terrorism? Of course they can be used for that purpose. Thanks a lot Senator Frist.

Here's a law that purports to end Internet gambling but will instead enable it, a law that is intended to make certain types of financial transactions harder to do but will ultimately make them easier, a law that says nothing about terrorism but will ultimately abet it, making us all less secure in the process.

There is, to my knowledge, no center for Al-Qaida hacking, nor is terrorism as an industry big enough to attract much third-party software development. But ally the interests of terrorists and Internet gamblers who all want to be paid, that's a $20 billion incentive to corrupt the world financial system -- an incentive that didn't exist before last week.

And what will be our institutional response to these obvious flaws when they come to light? More regulation of course! More scrutiny of financial transactions, not less. But as we've seen in recent years, this greater scrutiny often comes with lax or unequal enforcement, depending on your campaign contributions.

Once again, Congress is proposing to regulate something it ought not to -- something that in any practical sense is probably beyond its power. And the result will be only bad, not good. And Congress's response will probably be even more regulation, not less. And all this to push one man's presidential ambitions?

There ought to be a law against THAT.