Blog Post

The Network Computer Arrives…Finally!

Thank you Eric Schmidt for taking me down memory lane, to the heyday of another bubble, in another century. Today, at the launch of Chrome OS — a new Google (s goog) operating system for web-centric computing — Schmidt talked about 1997 when he (then at Sun Microsystems) and Larry Ellison and everyone else talked about the idea of a network computer.

Network computer, at the time, was defined as a stripped-down machine, with little or no moving parts, cheap processors and ample bandwidth to compute over the network. It was a computing equivalent of Silicon Valley’s Moby Dick. Larry Ellison, CEO of Oracle (s ORCL) was the proverbial Captain Ahab.

Larry “Captain Ahab” Ellison

In 1995, Larry Ellison talked about a network computer that would cost $500 a pop and would free the world from Microsoft (s msft). “A PC is a ridiculous device. What the world really wants is to plug into a wall to get electronic power, and plug in to get data,” said Ellison at the time.

His idea was a sound one, though at the time, a tad impractical. Oracle’s Network Computer subsidiary lost $175 million and was eventually spun out as Liberate Technologies, a company that transitioned to building software for set-top boxes.

Nevertheless, the idea had strong appeal for Silicon Valley’s power brokers, and they kept plowing dollars into the network computer. Marc Andreessen was an early champion. What was good for Ellison was good for IBM (s ibm) too. But the guy who was completely besotted by the idea of network computer was Eric Schmidt. It was Schmidt and his cohorts at Sun who came up with Java Station, a disaster if there was any. The Java-based network computer was underpowered when compared to the PCs of the day.

Have a Corona

The network computer hype had died down by 1998, but the Sun team hadn’t given up. They once again went back to the drawing board and started working on a network-business appliance, code-named Corona.

I first broke the story about Sun’s ultra-secret PC killer back in August 1999. Unlike an all-purpose consumer device, it was targeted at the enterprises and was pitched as “zero admin cost” machine. No need for expensive storage — just lean processors with ample bandwidth. It plug into the network and essentially brought up a “state” for folks to start working. You needed a special Java-enabled card that created the “state” on any device, regardless from where you were accessing the data. It was targeted at airlines, retailers and package transportation companies.

Sun eventually called it the Sun Ray, thus bringing a short-term revival of thin clients. Even a new version of Java Station was launched, to no avail. The network computer in early attempts failed because the devices were too expensive when compared to low-cost personal computers and bandwidth was still extremely expensive and the connections, even on corporate networks weren’t fast enough. The software and the experience just weren’t compelling enough, and there were limited use cases for the device.

The Past Is the Future

As Chrome OS was being launched, I couldn’t help but think about the similarities between the two products. The early customers for this new Chrome OS based computer (with no admin-costs) are American Airlines (s amr), Intercontinental Hotels and some retailers. Onstage, Schmidt said that he and his peers were talking about these very same problems — total cost of ownership, security and ease of use — when they were first discussing the network computers.

“We were right that the underlying problems really were a problem, but were wrong in understanding how complex the problems were,” he said. Of course, as time passed the networks got faster, the components got cheaper, but more importantly, the rise of AJAX and the evolution of the open source LAMP stack allowed the idea of web-based applications to blossom. And from there, evolved the idea of a web-based operation environment/system. “Chrome finally broke through architectural frameworks with respect to speed and security,” a beaming Schmidt said. “It is now finally possible to build powerful apps on top of a browser platform.”

Schmidt believes Chrome OS is the third viable real operating system, one that breaks from the past, and looks into a cloud-services centric future that would re-define the idea of what an OS should be.

With Chrome, Google has done a good job of executing on the idea of “network computer,” offering a suite of cloud services. In 2008, I wrote up a list of ten features that a good cloud client must have — Google has done one better with their prototype device, the CR-48.

“The long term vision of browser as an OS for web applications” said, Sundar Pichai, Google’s vice president of Product Management for Chrome said. Making browsers speedy and responsive to today’s web tasks, they have made it possible to do more inside the browser.

I was particularly impressed by the improvements they have made to JavaScript, a boon for web applications. Other extensions that tap into the hardware for graphical improvements will only going to make the browser more desktop-like. Pichai has to feel good — he is helping realize his boss’s vision for the network computer. Too bad, Schmidt and Ellison are not talking to each other anymore.

Related content from GigaOM Pro (sub. req.):

33 Responses to “The Network Computer Arrives…Finally!”

  1. John Connor

    On August 29 it gained self-awareness and the panicking operators, realizing the extent of its abilities, attempted to shut it down. Skynet perceived the attempt to deactivate it as an attack and came to the conclusion that all of humanity would attempt to destroy it. To defend itself, it determined that humanity should be exterminated.

    Before it could be deactivated, Skynet launched the US nuclear missiles under its command at Russia, to which Russia responded in kind by firing many of its nuclear missiles back at the United States and its allies. As a result of the nuclear exchange, over three billion humans were killed in just minutes. Governments around the world collapsed and anarchy descended.
    Isn’t anyone else worried about cloud computers becoming part of a larger network? The government already monitors every aspect of our daily lives. I kind of like the idea of unplugging and still having a local machine with MY data.

    • Mac OS and Linux really came from the same source tree, so they’re not really that different. If anything, I’d structure it ‘Microsoft,’ ‘UNIX,’ and ‘…’. UNIX is the basis for so many technologies that are used today, and to gloss over that is serious shortsightedness. Where do you think Android and Chromium have their roots?

    • What you might consider even more remarkable is that the original Sun Ray 1s that we shipped in 1999 are still quite usable, and indeed, can still be seen in offices on Sun campuses. We offered a 5-year warranty on the original devices because they had no moving parts, were low power, and had a ridiculous MTBF. So, it’s not good enough to consider just original purchase price when comparing against a PC. You have to consider amortized cost. If you want to improve your Sun Ray performance, you upgrade the server, not the appliance. We have indeed evolved the hardware over the years (usually because some chips we were using went end-of-life), and we’re now on the 3rd generation. We’ve added lots of interesting improvements along the way, including built-in VPN, lots of support for Windows access, and multi-media support.

      By the way, Sun was for awhile the biggest Sun Ray customer. Most of the user desktops in the company were replaced by Sun Rays. With “hotdesking”, which is what the smartcards were really used for, you could plug your card into any Sun Ray in a Sun office IN THE WORLD and it would cause the Sun Ray to connect to your existing session, even if it was halfway across the planet – with pretty good performance.

      I could write way more (correcting some more of the factual errors above, for instance), but that’s enough for now.

      (Composed on my home Sun Ray, with built-in VPN, connected to a server at the office…)

      (Opinions mine, not Oracle’s, etc.)

  2. And there was a whole generation before that… X terminals! Network Computing Devices (NCD), at least, made a business out of these for a while in the early 90s.

  3. Yes somehow I see this coming out and then 2 years later a big new feature of Chrome OS will be apps running locally!!!!

    2 years after that – local storage!!!!

  4. I imagine it’s still in the rotation. Bloomberg-TV has been doing an hour on Larry Ellison on their Game Changers program the past few weeks.

    You get to see him introduce the network computer all over again.

  5. PCs started because people wanted to get away from the main frame and terminal and have control of their data. Now we are being told that we don’t want control of our own data. Thirty years from now, the next big thing in computers will be moving off the cloud.

    • I think it’s going to less than 30 years when people will want to backup locally what they have in ‘the cloud’.
      It’ll be in less time than that that our electricity supply will no longer be dependable, and the server farms will be vulnerable.

  6. There is this little company called Wyse Technology that has been around for over twenty years who moved from computer terminals to thin clients (a.k.a. cloud clients) that you might not know.

    There’s times where web/cloud based computing is fine, but, call me old fashioned, I like holding my content close to me.

    That said, I’d love to evaluate the ChromeOS notebook to see what it can do. For time to market purposes and lower development costs, using an Intel Atom platform makes sense. But I’d really like to see an ARM-based version – cheaper, longer battery life, cooler, potentially just as fast. Frankly, if all the apps on it are web apps, who cares about x86?

  7. Om wrote: Of course, as time passed the networks got faster, the components got cheaper, but more importantly, the rise of AJAX and the evolution of the open source LAMP stack allowed the idea of web-based applications to blossom BECAUSE?.

    Was that sentence supposed to end like that?

  8. Glenn Stauffer

    It brings back memories of listening to Larry Ellison talk about the network computer at my first Oracle conference and using a SunRay as an email station at a conference a few years later. I ways thought it wax a great idea; and now with the application infrastructure that Google and others have built, finally!