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

<< [ Why Is Everyone So Excited? ]   |  Make Three Wishes and Call Me in the Morning  |   [ Please, May I Use My Own Money? ] >>

Weekly Column

Make Three Wishes and Call Me in the Morning: The Hidden Flaw in Sun's Jini Plan

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

In the world before Barbara Eden, jinnis were magical but malevolent creatures trapped in bottles and old oil lamps. Apparently, way back when, they washed ashore all the time. Rub the lamp and let them out, said Sheherezade, and they'd kill you rather than thank you. Only by tricking the jinni into returning to captivity could it be tamed, ransoming itself with three wishes. This week in San Francisco, Sun Microsystems unleashed its own jinni, and all three wishes appeared to be the same: Kill Microsoft, kill Microsoft, kill Microsoft.

Sun was in San Francisco pushing a new product called Jini, software built from Java and intended to bring easy-as-pie networking to everything from soda machines to supercomputers. Thirty-seven other companies stood with Sun as Jini licensees. It was a formidable display.

But let's understand something about these licensee lineups: They are pretty much meaningless. Look into the hearts and brains of those 37 company executives standing with Sun CEO Scott McNealy for the photo op, and you'll find more — and less — than you might expect. Some of those outfits were there because they love everything about Jini and are betting their companies on the networking of poker chips or pizza ovens. But some of those present have no real plans at all for Jini. They just want the publicity or to appear to be a high tech company, or to snub their noses at Microsoft. Some companies might be planning to offer Jini products, but this could be in a mix of many technologies, including those religiously opposed to both Java and Jini. Some companies like to hitchhike both sides of the road.

The only sure thing about those 37 is that not all of them will actually offer Jini technology in their products. This is just the way it is. Go to a Microsoft, Intel, Apple, IBM, Oracle — you name it — technology roadshow, and you'll see the same thing. Microsoft trotted out 35 OEM licensees for its old LAN Manager network operating system back in 1989. Only two of the 35 ever shipped product, and those two didn't make any money.

But this does not mean to say that Jini is any less than Sun claims it to be. Jini is a great idea brilliantly implemented. Like many Sun products, it takes a systems approach that amazes the personal computer crowd because while Sun really believes "the network is the computer," the rest of us have a hard time getting beyond "the computer is the computer."

The Jini concept is that anything you can stuff a Java virtual machine inside just ought to be networked. And since every Jini device is inherently networked, let's not make it such a big deal to get connected. Jini devices automatically connect themselves. They look for a network connection, and if it's there, they do the best job their tiny brains can to figure out how to talk over it, all without asking for IP addresses, subnet masks, or anything.

Even though it would seem the class of devices Jini is aimed at (garage door openers, door locks, e-mail machines) are stupid enough to need all the help they could get, it's this very weakness that makes transparent networking necessary. Jini devices had best figure out the network by themselves because they typically don't have the smarts to even know how to ask. So they have no choice but to make it look easy.

This is a big threat to Microsoft, of course, because Jini isn't PC-centric. Bill Gates is perfectly happy for your can opener to have networking capability, but he wants at the center of that kitchen network to be a PC running Windows. Sun's approach is to let the little Jini devices talk peer-to-peer when that makes sense. When there is a need for a central brain, it can be a Unix server across town.

Sun loves Jini, both because it is a solid validation of Java and because Microsoft doesn't own it. This, of course, bugs the hell out of Microsoft.

To combat Jini, Microsoft has Universal Plug-and-Play, which places a PC back in the center of things. UP&P is more like a slower extension of the PCI bus out across your home office to the stereo system and the dishwasher. Microsoft will trot out its own squad of 30-odd OEM licensees, some of them the same people who stood this week with Sun. And the number of those Microsoft licensees who will actually deliver product? Remembering the wonders of Microsoft at Work — the old plan to network copying machines and paper shredders — I have my doubts about UP&P.

And Jini has another great advantage: a really smart licensing policy. A Jini license costs 10 cents per unit or $250,000 per product, whichever is less, with no big up-front payments or even charges for development tools. McNealy really, really wants this thing to be a success. And why not? If Jini takes off, it puts Sun right in the heart of consumer devices. It offers the prospect of stealing much of the desktop networking market from Microsoft and Novell. (Sun doesn't trumpet this, but it's part of the dream.) And all this happens with no setup, no interface, no chance to voice the usual anti-Unix complaints that have kept Sun from success on mainstream desktops.

But there's a risk in this Jini stuff, a risk not all that different from that posed by Sheherezade's malevolent jinnis found on the beach. Jini might not be able to deliver the goods. In his Jini schtick, McNealy likes to talk about how Jini will put a tiny Web server in every device with a microprocessor. He uses the front brake pads in your car as an example. When brake pads get brains, McNealy dreams, those brains will use Jini to report their state of wear not only to the car through the car network, but also out of the car, onto the Internet. They'll shoot a signal all the way back to the brake parts factory, where the death of one pad will start the production of the next.

Heck of a dream, eh? But will it actually work? Here's the big news in this column: Sun has no idea if Jini will work as McNealy describes. This is an instance of marketing colliding with reality. Jini was designed for houses and small offices. The idea was that devices would sniff each other out and create ad hoc networks of a few or a few dozen devices. But the brake pad example Scott McNealy likes to use would imply a network of hundreds of millions — maybe billions — of nodes communicating over distances of thousands of miles, using both wired and wireless networks. In terms of sheer scale, this is beyond anything the Jini crew ever considered, and they literally can't guarantee it will work as advertised. They told me so.

This is not bad technology, not even bad marketing, and there is the very real possibility that the whole darned thing will work brilliantly. But scale effects are hard to predict. Big networks are nothing like little networks. So if it does work as McNealy says, there are engineers at Sun who will know their wishes have been granted.

Comments from the Tribe

Status: [CLOSED] read all comments (0)