Mad Dog 21/21: Tablet Vivant, Memories Mordant
Published: March 14, 2011
by Hesh Wiener
Hitachi, which once ate IBM's disk drive business, has regurgitated, selling its rotating storage operations to Western Digital. Meanwhile, outfits that track memory prices generally agree that DRAM has become a drug on the market. The only storage chips selling like hotcakes are the kind used to build flash memory devices. Even those current darlings of the electronics world may get whacked, pricewise, as the Korean kings of NAND segue to 30 nanometers. What's going on here? It's iPads on the client side and green blades on the server end of the wire, that's what.
A century ago, more or less, there were laws in modern countries, if you consider England a modern country, that attempted to ban hootchie-kootchie shows. But it turned out that, just as it is today, sex helped sell beer. So tavern owners studied the laws and figured out that the banned shows that involved what the cops and courts would call lewd acts. The banned acts in question involved dancing or at least motion. What if the lewdie nudies stood still? Generally speaking, that was legalish. The upshot in London (and elsewhere) was the evolution of shows that featured tableaux vivants.
In Paris, where a keen appreciation of nookie so often prevails, what Londoners did to beat the law and peddle warm beer was turned into stunning spectacle by the people who created joints like the Folies Bergere. The Folies presented tableaux that often gave it an edge over its similarly exotic rival, the windmill gin mill.
Tableau Vivant: Nearly a hundred years ago, Folies Bergere in Paris used tableau vivants every night to titillate customers.
The key lesson to be learned by staring at the sparsely dressed entertainers who stood still to please the crowds and mollify the law is that sometimes moving less is more moving. It is a lesson lately learned by disk drive makers who have seen some of their best markets decamp to the land of solid state, even though, as misguided analysts have pointed out, spinning hard drives are vastly less expensive than solid state alternatives. Cost is a big factor, but not the only one, not by a long shot; SSDs don't yet compete with spinning drives on cost.
A little over a year ago, marketing folk at Agigatech touted a presentation made by analysts from Objective Analysis that compared SSD price trends to hard drive price trends. OA also compared the pricing of flash to DRAM. Agigatech claimed SSDs were too expensive to bump off hard drives, at least for a little while, and that flash was cheap enough to serve as backup for faster but volatile DRAM. The OA presentation was made--and the Agagtech Web page written--before Apple's iPad turned the computer business upside down and before the everyone realized that skinny Mac laptops were never going to have hard drives anymore and that they would store their stuff only on SSDs.
Basically, tablet computers that use DRAM (and not all that much of it) and SSDs, but don't use spinning disks at all are all the rage. The leader is Apple's iPad but other tablets might yet give the original some stiff competition, the way Android phones, at least in aggregate, have more or less matched the iPhone's sales. However things play out, it looks like tablets are set to dominate (and enlarge) a consumer computer market formerly devoted to laptops. Regular laptops, most of them still using spinning hard drives, might hold on forever in the business market, particularly if the server companies like IBM keep misreading the market.
Hitachi, which makes servers and no longer wishes to make disks, seems to have a better understanding of what is going on than IBM. Hewlett-Packard or Dell might be a bit smarter than it looks so far. Oracle, which has the most outstanding opportunity to win or to lose right now, looks at least as out of touch as IBM. In fact, the rivalry for glass house business between IBM and Oracle may be distractive both in what could prove to be a fatal danse macabre. Both would be better off studying the tableau vivant stuff.
Here is what has been going on for a year and for many millions of client sales:
Clients have become dumb but skinny and physically resilient. They no longer need lots of DRAM, they have no room for hard drives, but they sure can use a lot of flash memory. Ordinary business applications that work well on desktops might not work so well with tablets. The businesses that will figure out how to rework their apps for tablets (and their dinky cousins, the smartphones) ought to be led by Amazon or one of its rivals. It's really simple: If Jeff Bezos cannot build a tablet Amazon that dominates the segment defined by the iPad, whoever beats Bezos gets to bury him. It's that simple.
Spinning down: Hard drives are much cheaper than SSDs but SSDs are nevertheless eating into hard drive sales.
Maybe Google has a chance. Maybe Facebook can evolve to become a peddler. Maybe Microsoft will enjoy a renaissance. Maybe Larry Ellison will either retire or give up pocket pool and find people with imagination to give his otherwise dead-ended company a future. And as I said, I think the last few years have given Michael Dell enough of a scare to put him in do-or-die mode, with the odds in his case favoring do.
What the server people need to do is what they have talked about doing since clients were called 3270s and 5250s. They need to make user-friendly stuff happen at their end of the wire and deliver a really fun and effective experience to the user carrying an iPad or iPhone or rough copies of one of those things. (If it can't be done then somebody has to make an iPad with real computing power and replicate PCs in slim packages.)
Memories are made of this: Flash long since beat DRAM in price but cannot match its performance.
In order to make all this work, the server side can't be something based on mainframes or mainframe thinking, which at its heart is about moving the fewest number of bytes possible up and down the network, even if that means driving the end user nuts. A server solution that works well with tablet clients might mean server farms set up so each end user has a whole virtual server. The kind of hardware that will be used to build these farms might not be giant boxes like mainframes or Watson. They might be stacks of teeny servers that use very little power, have SSD so there are no moving parts to fail, and use relatively lightweight CPUs and skinny DRAM to save on electricity. On these servers there may be databases still kept on spinning disks, but software and temporary storage and users' hot files probably will be moved out to the fence of blades that defines the outer edge of a server farm.
It this makes it sound as if Amazon is on the road to the future while IBM might not be--even if it has bolted cages of second tier servers to its mainframes and done a very good job of virtualizing the i--because none of its solutions deliver rich user support measured in watts not kilowatts, as clients move down toward milliwatts.
A very substantial portion of the booming market that is called electronic commerce is moving to the same gadgets that provide communications, navigation, information, and amusement. These machines don't have moving parts. No hard drives. No keyboards. No mice. They are not for everyone, but they are for so many someones that they are defining the near future of computing and of all the enterprise that depends on computing. The tableau vivant machine has made the old hootchie-kootchie obsolete. Now who would have ever imagined that?
Post this story to del.icio.us
Post this story to Digg
Post this story to Slashdot