The end of the NIC [LWN.net]

The "New Internet Computer" (NIC) was another one of Larry Ellison's Big Ideas: a low-price, hassle-free Linux-based computer which limited itself to Internet activity. It was essentially a browser with a bit of accompanying hardware. There was no hard disk; Linux would boot off a CDROM and what little data needed to be stored went into flash memory. At $200, it seemed like a cheap and easy way to get Linux onto desktops - and counter tops - where it had previously failed to go. LWN covered the NIC release back in July, 2000.

So much for that idea; three years later, as reported by ZDNet, the New Internet Computer Company is shutting down. Sales have been scarce, and the company was unable to come up with another round of financing. When you have a company that is not making money, there are really only a couple of choices: find an excuse to sue IBM, or shut down gracefully. NIC chose the latter path.

The Linux-based thin client (or "Internet appliance") product once looked like a good idea. Many people just want to play around on the net, and don't want to hassle with computers, software installations, drive failures, viruses, etc. Why not provide them with a simple box which handles this basic task and doesn't ask for any care and feeding?

The answer would seem to be that, when people want a computer, they want a real, general purpose computer. For the price of a NIC, it is possible to find real systems which can be customized, enhanced with additional software, and generally made more useful. People naturally shy away from a system which appears to offer reduced functionality or to be, in some way, crippled. This is, perhaps, especially true when people are looking at Linux systems, which ordinarily offer a greater degree of control than proprietary alternatives.

There may yet come a time when everything one might want is available as a web service, and users want little more than a display with a browser and a "buy" button. But, for now, it appears that the general purpose computer has not yet completed its run.

(Log in to post comments)

Thin-clients die again. Film at 11.

Posted Jun 5, 2003 6:47 UTC (Thu) by flewellyn (subscriber, #5047) [Link]

So yet another company made yet another try at the "thin client" or "network appliance" idea, and yet again it foundered on the simple fact that, really, it's cheaper to put together a general purpose computer than buy one of these specialized, crippled "network-only" machines. NCD X terminals, Sun Javastations, and now NICs seem to illustrate well the point that people don't want a simple "internet appliance".

Well, maybe corporate offices would value some kind of diskless machine that could be booted centrally, but that's easy to put together with stock hardware. All of the above examples were a bit too pricey, and too crippled; you could build a diskless PC-style machine for about the same cost, but have something that could be upgraded and reused as a real workstation or even a server later on.

So what makes companies keep trying to market specialized diskless machines? Managers may just be enthralled by bad ideas, loving the notion of succeeding at something that everyone else has failed at, but not considering that there may be a REASON for that failure.

Thin-clients die again. Film at 11.

Posted Jun 5, 2003 7:35 UTC (Thu) by dlang (subscriber, #313) [Link]

at one point in the past when drives were rather expensive building a machine without one made some sort of sense.

nowdays when a 40G drive is selling new for <$70 RETAIL there is no reason to make a diskless client.

however I think the real reason that they have failed is that microsoft doesn support the concept well (if at all) and the corporate desktop is still owned by them.

Many MIS departments would love to have more control over the machines, floppies are used more for loading software that will cause grief than for filesharing (walk into most companies nowdays and try to find a blank floppy disk, you may have quite a hunt on your hands), but until Linux is able to do the full job there really isn´t way to use such a machine.

besides which nowdays you can buy a complete machine for $200 so a stripped down machine will need to be quite a bit less (there´s not much room for a profit margin here) and the software to run on the machine and possibly the monitor or even the chair the employee is sitting on all cost more then the computer so why take the risk of useing a small volume supplier?

Thin-clients live again. Film at 11.

Posted Jun 5, 2003 13:18 UTC (Thu) by hazelsct (subscriber, #3659) [Link]

"nowdays when a 40G drive is selling new for <$70 RETAIL there is no reason to make a diskless client."

Perhaps not for home use, but in large corporate and government settings, the savings in administration can be enormous. To cite an old story, Largo, FL spends one third the IT budget of other cities (proportionally of course). by using thin X clients connected to Linux servers.

Of course it helps that they can buy these thin clients on EBay for five bucks each, because nobody else wants them. :-)

Thin-clients live again. Film at 11.

Posted Jun 5, 2003 19:19 UTC (Thu) by Baylink (subscriber, #755) [Link]

And it helps that they're *permitted* to do that...

Thanks; I was going to make much the same reply, but I forgot that the case in point
example was *2 miles* from where I'm sitting.

Thin-clients die again. Film at 11.

Posted Jun 5, 2003 15:58 UTC (Thu) by leandro (guest, #1460) [Link]

> when a 40G drive is selling new for <$70 RETAIL there is no reason to make a diskless client

Yes, there is: less moving parts, less software installation, less troubles. And a thin client is more a X terminal than a diskless client.

Thin-clients die again. Film at 11.

Posted Jun 5, 2003 11:15 UTC (Thu) by leandro (guest, #1460) [Link]

> NCD X terminals, Sun Javastations

They are still there, just with other names. MS is also promoting the concept with MS WTS, and the X terminal is much older and bigger than NCD. I have now a client wanting to implement a GNU/Linux server while reusing his 10 MS W9X clients as X terminals.

The end of the NIC

Posted Jun 5, 2003 8:47 UTC (Thu) by beejaybee (guest, #1581) [Link]

What people "want" is what they're told they want, through advertising. General-purpose computers are unreasonably complex for at least some large groups of users; NICs definitely do have a place, but are unlikely to be profitable until:

a) Microsoft and the bulk PC manufacturers are forced to stop misleading advertising implying that their products are as easy to use as a toaster;

b) network connectivity at hard disk access speeds (> 100 Mbps) is universally available at a generally affordable price (< $10 pm).

Until then, venture capitalists investing in NICs are pouring money down a drain.

The end of the NIC

Posted Jun 5, 2003 9:52 UTC (Thu) by mwilck (guest, #1966) [Link]

So let them pour it down the drain. Prerequisite b) may come to pass some time, but a) seems to be about as likely as "all the people living life in peace", to speak with John Lennon's words.

KNOPPIX, PXE, Bootable "thumb" drives.

Posted Jun 6, 2003 15:13 UTC (Fri) by AnswerGuy (subscriber, #1256) [Link]

We don't need NIC's underpowered wonders. Any old PC (like the Dell GX1
that I can get nearby for $180 each) and a KNOPPIX image gives you
all that the NIC did and comes with a 6GB drive to play with.

(I'm typing on one of my GX1's now).

Also most PCs made in the last couple of years include support for PXE
So you can run LTSP, PXES Thin Client, or the KNOPPIX Terminal Services
package (among many others) on them.

Basically any modern desktop PC with ethernet on the motherboard can be
used as a diskless thin client or xterm. They're cheap but can be
upgraded with additional RAM, hard disks etc with no worries.

For example this particular old Dell GX1 has a funny little SIIG
combination Firewire (IEE1394), USB2 and SATA (serial ATA) adapter in it
with an external IDE to Firewire drive adapter and an 80GB drive hanging
off the back. I could daisy chain more of those then I cared to off of
that. This same box was running a copy of KNOPPIX for a month (just one
session). I only rebooted it back under the test copy of RH9 on its
hard disk because my wife needed to borrow the CD.

The biggest reason for NIC to die is that there are too many other
alternatives that are simply better. The little Shuttle PCs available
at Fry's are close to the same price, retail and new. Used computers
are spilling out of warehouses and businesses everywhere you look.

The NIC was inconvenient (mail order) and limited (connecting a hard
drive was a hack if you wanted to do so).

Also I've heard that many newer motherboards can now boot from USB
devices like these little flash "thumb" or "cig@r" drives. 32Mb drives
cost about $30 and I have a 256Mb drive for less then $80; solid
state (no moving parts) and write protectable (with a small external
dip switch). Of course much large spinning drives are available for
about the same price; and flash media are unsuitable for frequent
writes. However, it means I can easily and inexpensively deploy
machines with few moving parts and with hardware write protection into
a thin client environment (so I might have to fiddle with them and put
one of the USB connectors inside the case with the thumb drive: big deal)
and they'll generate very little noise. (And they can boot and provide
basic functionality even if the network is down, the DHCP, tftp or NFS
servers aren't available, etc).

People want to own real computers

Posted Jun 6, 2003 17:10 UTC (Fri) by iabervon (subscriber, #722) [Link]

The real issue with a thin client is that it relies heavily on having a real computer somewhere if you actually want to get anything done. People will never be happy unless they actually own such a computer, and there hasn't been a push for consumer thin clients for consumer general purpose computers. Thin clients can't compete with general purpose computers, but they could suppliment them. Why not have the TV in the living room also function as a second display for the family computer? Most recent TVs probably have sufficient computing power to run an X server already; any DVR or DVD player certainly does.

I'm sure there are people who don't want a second computer so much as they want a second copy of their first computer. But nobody wants to have a NIC instead of the first computer.

The end of the NIC

Posted Jun 11, 2003 12:33 UTC (Wed) by rwmj (subscriber, #5474) [Link]

The answer would seem to be that, when people want a computer, they want a real, general purpose computer. For the price of a NIC, it is possible to find real systems which can be customized, enhanced with additional software, and generally made more useful.

I thought this until I recently went to Japan. There a lot of people have very basic-looking PDAs (they look more like pocket calculators, black and white only, and are *very* cheap). These machines are portable email clients - nothing else. They don't even show images, or have a web browser. Just for sending and receiving email wirelessly (over the mobile phone network I presume).

This is where NCs have failed. If an NC is really just a PC, then the hardware costs about the same as a PC, and because it's not sold in volume, the total costs is probably more than the cheap PC. Something which costs more than a PC but doesn't do nearly as much ... Hmmm. Wonder why those didn't sell. However, an NC which isn't really a PC, but is in fact more akin to a pocket calculator, can be used wirelessly, and costs no more than $50, there's something useful.

Rich.