3Com - a company built on ether 3Com's sale to Huawei Technologies and Bain Capital for $2.2bn marks the end of a company which pioneered Ethernet and network computing, and once owned Palm. 3Com was named for "computers, communication and compatibility" by one of its founders, Robert Metcalfe, who had come up with Ethernet as a networking technology while …
"We need a 3Com, willing to take risks and make mistakes, as if no-one takes the chances then nothing develops. Hopefully the new owners will maintain something of that pioneer spirit, they can certainly afford to."
And more Huawei too, who are Pioneering in the Western Greed Fields/Capital Markets. Bravo.
Their decline wasn't helped by them laying off loads of employees in the early 2000s, losing valuable employees with knowledge all in the pursuit of increasing the share price.
I worked there for a while and they laid off all of the web interface experts, the product I was working on would have shipped with really awful inconsistent graphics had I not redrawn them myself (luckily I grew up with DPaint and pixel based graphic apps).
USR modems? I'd keep quite about that one. Easiest fix ever for internet connection problems in dialup days - chuck the USR modem. Sportster? Well, I suppose you could make up a sport of "lob the shite modem" along the lines of shot putt or welly throwing.
Oh, and the AC power adaptors were great at blowing up anything that users mistakenly plugged onto the end. 30 - 40 volts AC for a modem? Telex maybe.
The collision management system adopted by 3Com over Ethernet may be an insult to intelligence, but the fact is that other protocols with more complex schemes and much more thought and money poured into them do not do better and have just as many collisions to manage.
So what is the use of paying more for the same ? Ethernet is robust and scalable, which is more than many other protocols can boast.
I worked at 3Com for a while. I did both Ethernet and Token Ring designs.
Ethernet blew Token Ring out of the water, on design simplicity, speed, reliability and cost. Token Ring was a nightmare to keep working. It may have worked fine at 4 megabits, using the special IBM cable, but when they tried to push it to 16 megabits over category 3 twisted pair, the technology wouldn't scale. It was an EMI nightmare, required proprietary MAC microcode on each node and never dropped below a 3-to-1 cost premium, compared to 10/100 megabit Ethernet interfaces.
There's a reason it's not around anymore. Token Ring had a narrow niche market, places where IBM convinced customers that reliability and determinism were worth paying for in performance and cost. That advantage disappeared with the advent of cheap 10/100 Ethernet switches, something else that 3Com was good at building.
3Com had its problems, but Ethernet was not one of them. For a long while, 3Com NIC cards were the premium quality choice. That advantage only faded when Intel got into the market.
Sorry for you, but all of the IT people I know cried when USR external modems were taken off the market, they were solid, eliable, and you could use them with any OS, I still have a Sportser 56K as abckup in case my router ever goes down!
I have no idea if you were using the grey knock off or something...
I would agree - the USR modems were pretty bad. In fact, when 33k6 came out, I went through four USR modems on three different machines trying to persuade them to communicate decently with the computer using standard Hayes commands.
In the end, I switched to generic internal cards (sadly losing the cool flashy lights) until the Winmodem became ubiquitous, then switched back to externals (diamond supra etc) until finally settling on ISDN before DSL came along.
The problem with everything faster than 28k8 seemed to be connection reliability (which is important for business). Best modem we ever had was a 14k4 Tornado external - worked every single time, fast & reliable.
All that having been said - US Robotics was a fabulously cool brand name to be using, even if the products were functional nor robotic.
I'm glad someone got some service out of USR/3Com modems - I'll freely admit that I cried when I was given Sportsters to use on projects (and not from joy). Perhaps the important factor is time - my Sportster mayhem memories are from about 1995 - that may be pre-3Com (doh)!
OK - what about 3Com 802.11 NICs? The hardware may have been brilliant - I'll never know, as the software was out-and-out tripe.
I am using one right now they never took them off the market as far as I can tell and this seems to work just fine and has for about three months hasn't failed yet. I think maybe you are all a bit poorly informed. Bought it new on line three months ago. Broadband cable is kind of high where I live, and no DSL.
The Sportster may have been a bit cheap and cack, but the Couriers were expensive and the real deal - if you wanted to make sure you could hold a connection across a really crappy phone line (possibly internationally) then you'd put the best hardware you could on the end - and that was a Courier...
Sportsters were mostly shit.
Now as a BBS sysop I can say that courier modems were good quality product, and the price reflected that.
Unlike every other modem I have ever used, a courier can be interrupted by someone picking up a handset on the modem line and after the handset is placed back on the hook, re-negotiate and continue what it was doing, instead of dripping the connection.
They were not the greatest modem (well maybe for a time when BBS systems used them for interconnects and would get an HST connection) but were good.
1)
You are wrong about ethernet. It's simplicity was a virtue, since it allowed cards to be designed and delivered years before competing protocols. The simplicity also allowed for the gradual deployment of a fundamental redesign of the basis of ethernet -- most ethernet today is switched, the world of collisions is a era ago.
2)
I'm surprised your history of 3Com neglected its "jump the shark" moment. 3Com totally abandoned the enterprise routing market. Abandon meaning no support whatsoever from the moment of the annoucement, in a market that prefers "end of shipping" notification several quarters beforehand and "end of life" several years hence. This abandonment killed 3Com sales of any other kit to enterprises: who'd buy an ethernet switch from a company that had demonstrated it would just walk away.
There is still money to be made from ethernet. But 3Com missed the core routing market and was hasn't been competitive in high-end switches for a decade. This leaves it dependent upon the same ethernet controllers every other supplier has access to, and thus only the ability to sell what the Cheap and Cheerfuls are selling.
I seem to have enough of them here, and they *seem* to have got the hang of it...
(oh yes - Courier good, Sportster, well, OK, in 1994 they were the cheapest game in town, but put them on the lines of a cable TV provider who whack up their attenuation to provide a good (read: loud) speech signal and spend a few weeks fielding calls from irate buyers until you can get the cable company to come and play ball. Happy days.)
This post has been deleted by its author
I agree with Glen Turner. How come no mention of the retreat?
I was working at a 3Com shop at the time and as part of the companies Y2K upgrades, we installed 3Com's CoreBuilder 7000 at most of the companies locations. OC3 speeds using LAN Emulation. I even invested the effort in becoming 'certified' on the CoreBuilder 7000. Horrible management interface. Especially compared to the CoreBuilder 5000 (I think the only thing the two product lines shared was a name) but those boxes are still delivering 155 Mbps and are rock solid. I hear that the company has finally manufactured a business case for removing them and moving to some Cisco shite solution but it remains to be seen if they can justify it yet.
Why? Because my boss at the time of the retreat convinced the hand wringers to buy spare equipment on E-Bay when the other organizations dumped there equipment in a panic. He bought enough spares for pennies to mitigate all risk and accommodate growth.
Hard to justify moving off of a solution that is solid and still exceeds capacity requirements.
Anyone actually deploy any CoreBuilder 9000's?
They got the Palm V on a plate, milked it & never did anything with it.
The classic. I know people still using it.
If they had had the vision they could have owned The Smartphone.
Instead they hardly spent a cent then spat it out when the pip got dry. Of course Palm's been lagging ever since.
Inevitable they went the same way. I've preached customers on the virtues of 3Com being worth the slight premium; but that was years ago.
Well, my 33.6k USR *internal* modem (circa 1995) didn't fail me until some freak motherboard short-circuit stuck it into "off-hook" mode. (Caused by a bad motherboard, in case you're wondering).
But it was a full modem anyway; I did have some problems under Linux but that was because I had set it in Plug and Play mode. I put the jumper back on to "COM3" and was happily using it with Linux, no drivers required.
When my dad gave away the last PC that the modem was on (1999), I was deprived of Linux dial-up until 2002, when one of that PC's successors came with a POS Conexant modem. However, that modem had Linux driver support, and so I was able to dialup again.
Currently, it is easier for me to use my *cell phone* as a modem than any of those POS "modems" that come with laptops.
I can agree that the USR sportsters were a tad ... variable quality. But the Courier v.Everything modems were synonymous with "the best modem moeny can buy", atleast they were until the combined ISDN and analogue versions popped up. I used a Courier on a FreeBSD box as my "home router" until January this year, when it was replaced by fiber optics and a c3640. The Courier was reliable under all conditions, period.
//Svein
...3Com's foray into VOIP, sort of: their NBX line of ethernet telephone switches. In the early 00s I was "certified" to install and configure these clunkers. They were ugly, unpredictable and built around VMWorks, the worst *NIX ever.
Ah, but 3Com's NICs are the best. I can't count how many "network" problems I've solved by disabling the integrated network interface (Intel chipset) on a PC and installing a 3C905B! I still keep a couple on hand. Just in case.
Considering Cisco blazed a path through the enterprise that early on 3Com was setup and could've simply been handed due to their existence prior to any real enterprise networking it's almost a wonder they didn't become what Cisco is today.
And I agree with Joe, about their handling of Palm was atrocious.
Was Nvidia. Let's face it: were the Intel integrated NICs worth crap until just a few years ago? Not really. They were software-based and had poor performance. Nvidia released a GOOD integrated NIC, and that forced all the other vendors to step up and upgrade their own products. After that, there just was no real need for PCI NICs, except in servers, and often not even there. I predicted 3Com's demise several years ago, and I'm not surprised to see them go. I still have a few 3Com 3C905C-TX cards laying around. I picked them up 5-7 years ago for $25 each.
What will be the next company to go down the crapper? Well, Cisco is a long way off from folding, but their decline is already evident. Other companies like D-Link are starting to produce products that are, in many ways, just as good. Why pay twice as much for a Cisco managed switch, when you can get one just as good from a competitor? One thing you do not see them competing with yet, though, are the routers themselves. I see Cisco slowly retreating into a router-only company (except for a few high-end backbone products), and then eventually disappearing as the rest of the industry catches up and produces cheaper, more user-friendly products. It will probably take another decade or more, but their end will come. You just don't design products like that anymore, and Cisco is unwilling to change.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
