ep0: 3c509 in test mode. Erase pencil mark! This means that someone has
scribbled with pencil in the test area on the card. Erase the pencil
mark and reboot. (This is not a joke).
It is the user’s responsibility (with help from the configuration program) to avoid configuring the boot PROM on the adapter in such a way that the system is not able to boot. If this does occur, you can manually jump the Test Via using a #2 pencil. The Test Via forces the adapter into test mode, which disables the boot PROM so that the adapter configuration program can be run. You must cover the designated area thoroughly with the pencil mark.
Test mode forces the adapter not to perform the Automatic Initialization sequence, which means the EEPROM is not read and the adapter is left in the following configuration:
■ Address Configuration register = 0000h
■ Resource Configuration register = 0000h
■ Product ID register = 0000h
Test mode also shortens the ID sequence to 8 bytes (first is still FFh and last is 69h) and forces it to the active state (that is, the adapter is active and will respond to I/O cycles at base address 0200h even without going through the ID sequence).
After you are done, thoroughly erase the pencil mark.
Good point, you can get sheets of plain white stickers from your local stationery store for about $0.10 a sheet, and cut them as needed, with scissors you can also get from that store.
Get a label maker, or depending what you're trying to do, a roll of whiteboard tape. Also "wet erase" markers exist, that hold up a bit better to being physically wiped.
What I really want is a network switch with an extra RGB LED per port that can indicate the VLAN.
Damn, too bad I'm in Mikrotik world. Not that I would have wanted to replace my switches regardless. Of course this has me left wondering if I just got the idea from something I had seen in passing.
I'd love to retrofit some solution with an LED strip/pcb (might have to custom fab to get the spacing right) and WLED. Except there isn't much room in 1U. (syncing configs would be a problem too but I could live with that)
Although I am going to build a new router (amd64/nixos), and it occurs to me that I should make some allowances for this (even though there are many fewer ports)
Wow, there's something so admirable about such a well-thought design that they even considered trusting their customers to be smart enough to be able to help themselves out of being in a pickle.
I worked at 3Com in the mid 1980's. We had a smart guy (Stanford grad) working in manufacturing that was doing numerical analysis on failure rates and product returns. His analysis showed that the 3c50x PC Ethernet cards almost never failed and built a case for offering a lifetime warranty. He calculated that it would cost the company almost nothing to offer the warranty.
They were high-quality cards and the lifetime warranty proved to be very popular.
In Thailand the first TP-Link distributor offered "lifetime forever" warranty. Even if it get lightning surge you can get a replacement. For that I believe they charged twice for the product. Many people think it was a really good deal for an unknown Chinese brand at the time.
Later on TP-Link themselves added more distributor and were not happy that they cannot dictate the pricing, and also the distributor did get undercut by the newer distributors who only offer standard warranty, so they part ways.
Even now almost 10 years after that they probably will honor my TP-Link warranty but I'll get a Totolink instead
That’s interesting, thank you for they tidbit! Do you have any other stories? Personally I loved the pcmcia cards but mostly because drivers were included from 98se, so in the early 00’s my first job at a of repair shop that was an easy way to get a computer online/networked.
Outside that, I think I missed everything though I have memories of another company that merged into 3x in, xircom? Before they were acquired again.
It is, apparently, though I don't think I've ever used it under Linux. The driver I wrote ran under pSOS, an old real-time OS we were using for a building access control product. pSOS was obsoleted by competitor VxWorks about a quarter century ago.
The fun thing about writing an Ethernet driver back then was that all of my knowledge of networking came from downloading RFCs, reading TCP/IP Illustrated and studying 3Com's 3C5x9 technical reference. Writing driver code was like solving an elaborate puzzle.
The 3C509, and the following 3C905, just worked. Back before Windows 95 and plug-and-play, getting hardware to work took actual effort. But those network cards worked with pretty much everything. Then when Linux started to become a thing, once again, they just worked.
Lifetime warranties are sort of a scam, mainly because they mean different things to each side of the transaction or perhaps even nothing. The customer thinks "my lifetime". The manufacturer, if honest thinks "the lifetime of us manufacturing the product", if dishonest "the lifetime of the product, when it breaks the warranty is over"
Does the manufacturer have to define what "lifetime" means? Perhaps buried in the small print somewhere.
I didn't see the thread about the lifetime warranty but for 3com stuff - means there is still firmware being released and I can call HP enterprise and get a new switch even though my current gear is old enough to legally drink. I had a bunk dell dock that kept causing arp table corruption and they were able to help me with email support in 24 hours. Pretty awesome actually.
When I was young we had a bunch of their etherstack hubs and 3c509 cards - it was fun being able to watch the lights blink showing a packet coming from one port and go cascading down the stack...until you'd have a packet storm.
Lifetime is defined in US law. Most people don't know what the law says. I don't, And I can't be bothered to look it up. But there is a definition in US law.
HP themselves sold Aruba networking equipment with lifetime warranties and then just decided they wouldn’t honor it anymore. So a warranty change after a company acquisition sounds even more plausible in comparison.
I guess they decided no one would take them to court, and if someone did they'd just refund the value of the item (which presumably is low these days).
Which would imply such lifetime guarantee promises are worth very little.
I have a very personal connection with 3com, though it was in the late 90s.
Ron Crane which was 3com's 4th employee and designed the 3C100, 3com's first ethernet product. Many years later he hired me for my first real job of out school and brought me to the bay area. I then got to work side by side with him and David Boggs (https://en.wikipedia.org/wiki/David_Boggs). At the time I never realized how special working with these two were...wasn't every company in the bay area like this?
I stayed in tough with Ron Crane until his death in 2017. He was both brilliant and incredibly sweet. He was so proud to take me behind the scenes at the Computer History Museum to show some work he'd been going.
One day, Ron (a hardware engineer from Xerox) gave a talk about the theoretical performance of an Ethernet file transfer. He calculated the end-to-end latency and data-rate including disk latency, network latency, the number of clock cycles necessary to send and receive the data, etc.
He had (overhead projector) slides that showed all his assumptions and calculations.
Finally he displayed a slide with a graph showing this theoretical maximum, the actual performance of the 3Com network stack, and (IIRC) the Novell stack. As I recall, actual performance was about an order-of-magnitude lower than the theoretical max.
His point wasn't that our software division should be able to achieve the theoretical max. It was that if we (for example) got halfway there, we would know that Novell would never ship software that was twice as fast as ours.
He was very nice and scholarly about it and did not mean offense. I and many others weren't offended. But a surprising number of people were.
The Computer History Museum has a nice two-part oral history of Bob Metcalf (3Com founder and co-inventor of Ethernet) back in 2006: https://youtube.com/watch?v=zKz07DdaKzw
Metcalfe has a reputation for being brash and outspoken and it clearly shows in the interview. I think it served him well in business and later in life as a pundit. I remember looking forward to his columns in InfoWorld, though I didn't necessarily give much stock to his predictions.
He talks a little bit about the downfall of 3Com in the interview, though he begs-off of talking about things that happened there after he left.
I very much enjoyed the trivia and minutiae about the creation of Ethernet and the Digital / Intel / Xerox relationship that ultimately resulted in it being standardized.
Basically, it was a long listen but there was some good stuff in there.
It was great to work with Metcalfe. I have two funny stories to share.
I was on the Macintosh network os client team. For whatever reason (maybe he was annoyed that Apple copied so much from Xerox) Metcalfe didn't like Macs. In this era Macs used Apple's proprietary AppleTalk/LocalTalk network. We did a demo for 3Com execs using a large external monitor connected via a single black coax cable that looked just like thin-ethernet. He got excited for a moment and came right over to look at the Mac and was very disappointed to see the cable attached to a display! (3Com was working on Ethernet cards for Mac at the time but they weren't ready at the time of this demo.)
For those too young to remember: there was an SF Bay Area baseball park known as Candlestick Park that was briefly branded "3Com Park" -- to the great annoyance of SF Giants fans. During one interview he as asked if in the early days of 3Com he would have done anything differently. He said, "I would have named the company 'Candlestick'!"
I remember those days, with Computer Currents' "Year of the LAN" parody predictions repeated year after year. Macs with PhoneNET definitely made it easier and cheaper in the beginning.
Do you remember which switch models it did happen to you?
I remember that SuperStack 2700 ethernet switches with ATM LEC (lan emulation client) card it did happen under prolonged heavy load heat accumulated with then caused sudden reboots. Occasionally these events switch was as factory reset was done, none of configuration made was left.
First with ATM LANE with ethernet STP (802.3D) was flimsy and buggy, so it took a while when that was fixed without STP spurious loops broadcast storms did quite easily overload edge switches and cause those sudden reboots.
We had to keep minimal ethernet management part configuration printed next to switches so that first entering primary network settings we could TFTP-load production config and get switch back to production quicker. But long without OOB management (serial port & remote telnet) it was always was hurry to get on site fixing locally these events.
I wasn't earning those occasions much after we got most often issues caused switches replaced with another other brand with did not suffer these problems.
> They sold a managed switch that couldn't save its configuration
I don't hate that idea. I wish I could set up our Cisco switches to do that, so on startup they pull down their config via ZTP and never ever ever save local changes.
That'll stop people logging in and making local changes and expecting everyone else to somehow just know about their tampering.
>That'll stop people logging in and making local changes and expecting everyone else to somehow just know about their tampering.
How about changing "enable secret" password and not sharing it with those who you do not approve making changes?
Or even setting up TACACS+ or RADIUS service and getting SSO to network gear that way. TACACS+ even free versions by default as an AAA that last A accounting makes you able to log all logins and commands being executecuted IOS cli.
Also, you can set up a management server (station) which collects all configs, checks in a db or VCS (RCS, CSV, git) which you can then build on changes monitoring, revert back configs as far as you have saved config versions. Like I had while working over 20 years backlog about half thousand switchers and other gear. RANCiD can do it for cheap, if you don't like to buy commercial solution which also can do similar things. Oh, OpenNMS and FreeNMS probably can do this also if you prefer getting those. I didn't use those as we built from mid 90's own systems and didn't bother changing at least when I was still working.
If you still need to allow other users login let other commands that enable level requires, you can use a "parser view group-name" and pick what additional commands you let them also access from exec enable level (15) without letting then become full enable 15 level users.
ps. Cisco IOS has long had service which autoloads config from TFTP server, but that isn't perhaps what you are looking for. It's mostly intended getting initial config uploaded to system enabling remote management over telnet and further making proper config automation easier.
> How about changing "enable secret" password and not sharing it with those who you do not approve making changes?
Ideally the management level wouldn't be on the tools, yes.
> Or even setting up TACACS+ or RADIUS service and getting SSO to network gear that way. TACACS+ even free versions by default as an AAA that last A accounting makes you able to log all logins and commands being executecuted IOS cli.
Sure would be nice to keep the management level out of RBAC
> Also, you can set up a management server (station) which collects all configs, checks in a db or VCS (RCS, CSV, git) which you can then build on changes monitoring, revert back configs as far as you have saved config versions
"I predict the Internet, which only just recently got this section here in InfoWorld, will soon go spectacularly supernova and in 1996 catastrophically collapse."
Keep in mind he co-invented Ethernet, so he knew what he was talking about!
Here's hoping we'll be able to look back with equal amusement at the AI doom predictions.
Prior to the time period covered in the article, 3Com had a major disaster in Software. We were losing badly in competition with Novell in the Network Operating System category (though I and others would argue we had a technically superior product) and decided that we would partner with Microsoft and overpower Novell that way. Big mistake.
I was only a consultant at 3Com during this time, so I may have some of the details wrong, but 3Com essentially bet the success of its software division on the partnership. We built a product called 3+Open that was based on OS/2 Lan Manager. The product flopped. But unfortunately 3Com had agreed to multi-million-dollar, fixed annual license payments to Microsoft for OS/2. (Which may have been the right choice had the product been successful.)
To get out of the license payments, 3Com gave all its LAN software IP to Microsoft and exited that business. Microsoft used that technology in Windows NT Lan Manager and essentially destroyed Novell.
If you want details, the Bill Gates Biography "Hard Drive: Bill Gates and the Making of the Microsoft Empire" tells the story of that failed partnership.
I worked for a while doing pre sales support for a company that sold 3Com products. 3+Share and 3+Open were pretty solid but mostly people just bought Netware. The XNS network protocol couldn’t be routed, only bridged, our biggest customer had 3Station disk less PCs and a huge number of 3Servers on a huge national network that gave them headaches managing due to broadcast traffic travelling over slow links. It was a fascinating problem to have in the era before properly smart gear.
Which in some beta versions was called "LAN Manager for Microsoft Windows NT", so they intended to keep the branding at one point. This carried through to release in a few ways, there's still a library called ntlanman.dll in system32, the legacy authentication scheme you should never use on windows is NTML authentication, and on NT 3.x/4.x the server SQU wallpaper was lanmannt.bmp / lanma256.bmp.
Not sure if there's much direct carry over code in NT 3.1's networking code from OS/2 though.
Recently saw 2 different network closets one filled with old 3com and HPE switches and the other one was HPE and Aruba. Basically the same switches, basic 24 port 1GbE. Just feeding a bunch of cameras with poe. Surprised the resilience of 3com hardware. Surprised how little changed in-between these generations.
First time i saw 3com's they were already ancient. One of the switches had a coax cable that connected to a hut that was too far away for Ethernet, wifi was not permitted, no one wanted to pull an optic, so we basically replaced everything around that hut an that switch with Mikrotik and just let it to chug along way past it's expiration date.
I worked at Convergent Technologies at the time, and our CEO went on a vacation cruise and met Metcalf. They decided over drinks with little umbrellas in them, to merge our companies. We were the largest single customer of Intel processors at the time(the PC market was very fragmented).
But the stocks jiggled and the buyout plan was postponed when valuations became uneven, and it never happened.
The 3c509 was the card we slapped into computers when the cheap NE2000 clones wouldn't work. Always did.
Years later I had a EtherLink III MCA 3c529 at home. I think it was one of two Microchannel cards they offered. Corporate sites usually had the IBM Lan Adapter/A since they were typically True Blue.
I also had a Office Connect ISDN LAN Modem 3c891 during the year I had working ISDN through Bell South. It supported channel bonding so I could get 128kbps, and it would automatically dial my ISP [0] whenever a machine on the local network wanted to go on the internet. Truly useful.
network cards mostly died when it got integrated into the motherboard - same with soundcards and even graphic cards unless you wanted pro quality sound, extra networking features or a gamer. For everybody else, the basic integrated options were enough. I grew up in that era and not once did I look up what the 3com card offered over another networking card other than 10/100. Just wanted it to work. I'm sure some people are more discerning but not much.
There were some “interesting” issues with a lot of the cheap hardware for a long time though. A classic example was the WinModem. A fast modem for cheap? Sure! But it was cheap because it outsourced a lot of its processing to the CPU, so if your system was loaded up (in the day when CPUs were almost all single-core) meant there wasn’t enough left over to process inbound data in realtime which meant dropped packets and retries and horrendous lag and throughput.
Standard wisdom was to use an Intel EtherExpress Pro card for a long time after onboard Ethernet was available. Lots of server hardware had unused builtin NICs until they got decent.
The separate cards may have died, but the same companies, or different ones, are making the chips that go on the motherboard. I think mine has a 2.5G Marvell and a 10G Intel chip.
"The RealTek 8139 PCI NIC redefines the meaning of 'low end.' This is
probably the worst PCI ethernet controller ever made...8139 supports bus-master
DMA, but it has a terrible interface that nullifies any performance
gains that bus-master DMA usually offers.
For transmission, the chip offers a series of four TX descriptor
registers. Each transmit frame must be in a contiguous buffer, aligned
on a longword (32-bit) boundary. This means we almost always have to
do mbuf copies in order to transmit a frame, except in the unlikely
case where a) the packet fits into a single mbuf, and b) the packet
is 32-bit aligned within the mbuf's data area. The presence of only
four descriptor registers means that we can never have more than four
packets queued for transmission at any one time.
Reception is not much better. The driver has to allocate a single large
buffer area (up to 64K in size) into which the chip will DMA received
frames. Because we don't know where within this region received packets
will begin or end, we have no choice but to copy data from the buffer
area into mbufs in order to pass the packets up to the higher protocol
levels.
It's impossible given this rotten design to really achieve decent
performance at 100Mbps, unless you happen to have a 400Mhz PII or
some equally overmuscled CPU to drive it."
tldr tldr: it didnt matter past 2001 since slowest low end CPUs were cheap enough to nullify all the potential gains from smart good 100Mbit NICs
No amount of software would have helped. It had limitations as to where it could fetch data from and where it put received packets, both requiring copying all network data around. Plus shallow TX queue resulting in higher interrupt load. Afaik 3Coms had proper descriptor rings DMA Master engine allowing zero copy and deep queues.
You couldnt saturate 100Mbit link with weak CPU on Realtek while 3Com did all the data shuffling for free. It didnt matter when browsing internet on contemporary slow links, and stopped being noticeable when copying data over Lan on >400MHz CPUs.
Did anything stop you from processing the packets in the place they were received and putting the data you want to send in the place it could be transmitted from?
They made good stuff. Then the cheap stuff got good enough for 90 percent of the use cases, then they bought into another sector where exactly the same thing happened again, except it was also a completely dead end. I don’t think any peripheral-card-only makers survived except Nvidia
I installed a bunch of 3Server network servers in the late 80s-mid 90s. They were pretty darned solid. Tech support was an actual human, I knew his name (though have forgotten it now), and we actually went out to lunch one time, as my company was in Mt View and they were in Santa Clara, I think.
I used to have a 905 on my router's external interface, until one week I had to do a modem swap. A couple days after the guy left, I went back in to do some cable cleanup... and when I plugged the cable back in, the surface-mounted jack sheared right off the pcb. While I could have soldered it back on, I'm surprised most cards didn't meet the same fate early on.
(2022) Yeah.. I know, it says 2026 at the top, but it was posted here in 2025, and you can see it's older. No idea why the author keeps ever-greening (the create date should at least reflect the original post?)
Yes, bought TippingPoint[1], one of the first NGIPS makers. That were one of first very capable L2 IPS to market. I did have opportunity to test one for a while, but did not get funding for it approved.
3Com was also one of the early ATM gear vendors with CELLplex 7000 ATM switch and SuperStack 2700 ethernet switches with LEC. I would say mostly[2] reliable and working gear few years before Cisco LS1010's ATM switches, Catalyst 5000 and first 3000 series ethernet switches after acquiring Kalpana.
2) SuperStack switches had bad habit suddenly reboot under prolong3ed heavy load and perhaps heat problems too, which then also occasionally did factory reset and lose configuration. And the menu based configuration with binary only tftp upload/download configs when better alternatives came available these were replaced any location where that ever happened causing issues.
I can't quite remember if the 3c509, 595 or 905 was the fastest and best. I think 905. I created a boot disc for these and a few other NICs that featured on Novell's Cool Solutions.
I remember also original EtherLink 3C501 and 3C505 with 3+Share yay!
Both were quite exiting and beautiful looking new fancy cards early -85 when I first had them on my desk. But in reality 501 was a dumb as a dodo and 505 with its dedicated 80186 (co)processor for 3+Share fileserver intended use not much better.
Here are links to pictures of those wonders we got -80 and cursed them later :)
I bought a 3c595 FastEtherlink card from CompUSA when I was in college because my dorm had ethernet and I was running FreeBSD and I thought tech specs were the be-all end-all.
It was fast, but not noticeably faster than any other cards my friends had. Except I spent way too much on mine.
3c509 was 10base-T ISA, so it's the slowest. 3c595 and 3c905 are 10/100 PCI cards, but the 905 was newer and more efficient (better bus mastering, more CPU offload)
This is a very, very common story for many tech corporations. They find a place then fail to adapt. Adaptec (ha ha) is another example that could not adapt.
This inability to adapt is inherent in the corporate business model. Look at Apple. They have a chicken that lays golden eggs. Then the chicken dies. Oops. So
they invest in speculative funding of massive proportions. Just ask AI and it will tell you.
Maybe there needs to be some other form of business that can do better?
Didnt Adaptec invest heavily in Fibre Channel? From what I remember from Computer History Museum CEO interview they went all in on FC and didnt see a point investing in InfinityBand as they saw it as also ran.
Article is a bit wrong, it's HPE that owns it now, not HP. Comware switches are still actively developed and sold. It's very solid, deployed in places you wouldn't expect where it just runs for years without issue.
What happened to them? They never really diversified out of network cards, their cards were very very good, but for most people paying 3x as much wasn't worth it.
They split off their Palm (PalmPilot) division in 2000 to focus on Network equipment. The the .com bust happened.
Palm/Handspring/Treo could have been the iPhone, if they'd been better managed.
Apple was also powerful enough (and Cingular at the time, weak enough) that they could dictate the terms and do whatever they wanted. Anyone else would have had their vision ruined by the cellular carriers.
I had a Treo fairly early on. Leave aside the physical keyboard. The iPhone had a whole app and media infrastructure built around it. Could Palm have gone there? Possibly. But that would have been a big step from where they were in the mid/late 2000s.
but the windows phone OS was better, and the hardware very good. They just couldn't get their shit together with the app ecosystem; even paying us devs huge premiums to write apps wasn't enough.
Nope. Innovation doesn't happen that way. Once they took on the keyboard path, it's nearly impossible to innovate into something that's not. Even if a team is innovative enough to come up with it, top level will kill it off
3com was the very first investment I ever bought…I bought my shares on Thursday October 15th, 1987….those that are old enough to remember will know how that turned out.
I can maybe do you one better. I think my first investment was 140 shares of LNUX at F&F IPO pricing of $30/share, purchased Dec, 1999. It opened at $299/share the day of the IPO. Should've sold that day. Eventually sold it all in 2004 for $2.80/share.
A fun article but it is incomplete. It misses out on the company's height.
In the early days it has its own server OS, 3Com 3+Share. It was a multitasking file server OS and it was based on MS-DOS. Yes really. An internally multitasking server stack on DOS. I became expert in it in my first job and my 3rd job leant heavily on that and my skills.
It used the MS LAN Manager stack over NetBEUI.
It inherited DOS limitations like 32MB partitions. On PC servers everything had to for into 640kB of RAM and natively it could not use XMS, only EMS.
3Com offered a range of dedicated network file server boxes, the 3Server. The first was an 80186 and later the 3Server3 with a 386.
Then it got big into OS/2 and released 3Com 3+Open. The multitasking stuff obviated by OS/2 1.0's own built in multitasking but now it was multiprotocol and the server could run other stuff too.
The 3Server had no display, no port for one, no keyboard connector. It could only be administered over the network. It did have a little 1-line LCD display for showing a status message, accessible via a DOS device driver for showing custom messages. A physical rotating dial let you choose a station number at hardware level.
Because there was no display, that part of the 80186 memory map could be used for RAM, so a 3Server could give 800-900kB of RAM to DOS. A selling point: it could have 25% more memory than a PC and was arguably easier to configure.
The 3Server3 had for the 1980s very high memory limits, but we didn't sell many.
3Com's DOS apps had their own pre-CUA text-mode windowing and menuing system, 3+Menus, with a consistent UI and navigation keys, before Windows 2.0. Fill in fields in any order, Enter to accept, Ctrl+Enter to accept the whole dialog box, close it (with an animated zoom-out effect), and return to the previous level. This applied to all 3Com software and meant very fast navigation. Last seen in the DOS setup tool for the 3C589 ISA NIC.
As well as file and print, it bundled an email server client, which almost none of my clients ever used, and it also could act as a network gateway, with tools for WAN linking of networks into a big MAN, which I also never used.
I remember putting 3+Share on an IBM PS/2 Model 80 with the biggest hard disk I'd ever seen at that time: 330MB, on an ESDI interface. I had to partition it as drives C:, D:, E:, F:, G:, H:, I:, J:, K: and L:. All 32MB except for an 11MB L: drive.
NOS on C:, main network app on the next letter, data on the next letter, home directories on the next, print spools on the next... it tool some ingenuity to find ways to use them all on DOS 3.3.
There was only one supported big-partition tool, VFeature Deluxe from Golden Bow Systems, and none of my clients were willing to pay for it. GBS did various DOS addons: a software disk cache, VCache; a defragmenter, VOpt; and VFeature, for disk partitions over 32MB.
I even tried installing it on Compaq DOS 3.31 which had big FAT16 partition support using the same system that MS-DOS 4 later adopted, but 3+Share could not see the big drives.
>The first thing that happened was 3Com walked away from its enterprise networking equipment business. Admittedly Cisco was hard to compete with. But 3Com was the only brand succeeding at doing it.
I can't really agree with that because 3com never had a serious router product line. They had LAN switching equipment and NICs. There was nothing like a whole product line of routers that could take a full BGP table in the 1999-2000-2001 era, and certainly nothing like medium to large carrier routers. I'm talking about around the same era when the Juniper M40 was released to compete with Cisco in that market. Or a Cisco 12008. Anything broadly along those lines. 3Com was far behind in advanced IP routing and things that would serve as like a metro aggregation router or core router.
A few lessons from my early IT life, all by way of 3Com...
A strange OSPF problem with 3com NetBuilder routers of which nobody could get to the bottom of. Based in a team that maintained the 3Com relationship with this arm of the (UK) civil service, who solely used 3Com networking products (LAN & WAN), we duly rang our contact and described the problem (and attempted fixes)... "You should change the PSU(s)" he said (eh? it's an OSPF problem?!?!?). Apparently they had some sort of (RF?) issue(s) with certain NetBuilder PSUs that had been known (note: undocumented, at least not publicly) to cause "strange" issues (it did resolve the problem).
Lesson 1: Sometimes it's the thing that wouldn't even occur to you that are the root cause.
Having ordered lots (>£1mn) of 3Com enterprise networking equipment (from LAN switches to their CoreBuiler 9000 chassis), and stuck it all in a warehouse, in preparation for a country-wide network upgrade - 3Com shortly thereafter (and seemingly out of the blue) announced their complete withdrawal from the enterprise network market. 3Com's messaging, as I recall, was pretty much - once the existing stock is gone there will be no more, and hardware support would last 2 years and after that - tough! Needless to say this caused quite some consternation with civil service execs, of which I assume invited 3com to a meeting (without biscuits) to discuss the situation. 3Com eventually partially walked back on said withdrawal (continuing to manufacture/sell smaller variants - and card/PSU/etc compatible - of the CoreBuilder 9000) although this wasn't enough to un-sour the relationship.
Lesson 2: Maybe this isn't the best way to treat your customers, and no matter how large the vendor (or how much you spend with them) - this may happen to you.
In one of 3Com's offices (Partington, a delightful place) - if you parked near the edge of the, fenced off, office car park - you may return to your car to find a brick embedded in it as a gift from the one of the locals. I assume much merriment was had by the cheeky scamps when tossing bricks over the fence at 3Com employee's cars.
Lesson 3: Get to this 3Com office early and park in the center of the car park.
Sat with a (TUPEd ex-civil servant, and hence un-sackable) senior network specialist who was updating the network monitoring for all devices in a specific region. This involved refreshing (via SNMP) the interface list of all the routers/switches/etc. During this process, which was basically click refresh - move onto the next device - and then revisit the original device after a while to confirm it had completed, one of the edge routers dropped offline (and didn't return). No problem - traffic would route (and bridge! IPX FTW!) via the backup router/link. The senior network specialist was quite methodical in his polling, moving from one (now offline) edge and around the ring of devices before eventually arriving at the final (note: backup - and now only - route) edge router (NetBuilder). Prior to reaching the final router it had dawned on my little (junior, green) brain that maybe he had triggered a (documented) NetBuilder bug that would cause the router to hang - needing a physical reboot - if SNMP walking a device with an FDDI card (which these routers had). Having voiced my theory, and in doing so suggesting that polling the final router may not be the wisest course of action, he confirmed that it was his job to poll the routers and that is what he was damn well going to do. The backup router dropped offline. 100s/1000s of civil servants now unable to do their job (that required mainframe access that resided beyond the edge routers) and needing physical intervention (i.e. ~1 hour) to resolve. Based on previous observations, I assume productivity for those users dropped by a single-digit percentage within that dark hour.
Lesson 4 (one of many such lessons reinforcing this view): 3Com's enterprise networking software sure has some strange bugs. Also, welcome to the public sector (or public sector adjacent).
Given the various bugs observed during my time in the team - I didn't hold the quality of 3Com's enterprise networking products' software in very high regard. Such low opinions would only be surpassed in later years by Fortinet's security/networking products' software releases.
I am sure that the built-in Intel Ethernet ports on most motherboards have made Network Cards obsolete at least for the Home Market. When I got a 2Wire DSL modem, they threw in an Intel Ethernet card with it for free. Now I use WIFI 6 on my Wifi router and PC, and it is fast enough for me.
I remember when USRobotics was giving away free 56Kbps modems to BBSes if the SYSOP listed that they had a USRobotics modem and kept the BBS up for 24 hours a day.
> I am sure that the built-in Intel Ethernet ports on most motherboards have made Network Cards obsolete at least for the Home Market.
Someone is still making a NIC chip, they're just selling it to the motherboard OEM, instead of selling an add-in board to the customer (or selling a chip to the add-in maker). In a way, this is more sales, because desktops tend to come with a wired nic and a wireless nic, and so do good laptops.
You are understating how integrated this stuff is. If you make an Intel client motherboard, all you need to add for the network is the phy, jack and some passives.
I haven't really been following intel boards. AM4/AM5 chipsets (and cpus) don't have integrated MACs. I didn't realize intel was putting them in for client parts (Xeon-D had them, but Xeon-D is weird :p)
My favorite 3Com-related thing: https://web.archive.org/web/20210421224904/https://nixdoc.ne...
It's the E1 pad between the barcode and the empty BIOS socket: https://en.wikipedia.org/wiki/3Com_3c509#/media/File:3Com_3C...Details (pg 88): https://www.ardent-tool.com/NIC/3c5x9b_Technical_Reference.p...
----------------------------------------
It is the user’s responsibility (with help from the configuration program) to avoid configuring the boot PROM on the adapter in such a way that the system is not able to boot. If this does occur, you can manually jump the Test Via using a #2 pencil. The Test Via forces the adapter into test mode, which disables the boot PROM so that the adapter configuration program can be run. You must cover the designated area thoroughly with the pencil mark.
Test mode forces the adapter not to perform the Automatic Initialization sequence, which means the EEPROM is not read and the adapter is left in the following configuration:
■ Address Configuration register = 0000h
■ Resource Configuration register = 0000h
■ Product ID register = 0000h
Test mode also shortens the ID sequence to 8 bytes (first is still FFh and last is 69h) and forces it to the active state (that is, the adapter is active and will respond to I/O cycles at base address 0200h even without going through the ID sequence).
After you are done, thoroughly erase the pencil mark.
----------------------------------------
I have two in use 3com 24 port switches. HP still is releasing firmware for them 20 some years later and still are under warranty.
Shows once again that electronics should have areas where users can write things. It's ridiculous that most USB drives don't have this.
You can, just use a permanent marker
But it's always a hack (a hack is defined as: using a thing in a way it was not intended to be used)
And some USB drives are all black, so you have to come up with something else.
So, no, "just use a permanent marker" doesn't solve this issue at its core.
Writing on things with a permanent marker is not a hack. It is quite literally the purpose of a permanent marker.
To write on black surfaces, there are white coatings you can apply first. You can try correction fluid, or white paint.
Floppy disks always came with stick-on labels, though. USB drives almost never do.
Good point, you can get sheets of plain white stickers from your local stationery store for about $0.10 a sheet, and cut them as needed, with scissors you can also get from that store.
PNY made these: https://www.computerworld.com/article/1549823/back-to-school... https://images.techhive.com/images/idge/imported/article/ctw...
of course they were too practical and working to get traction.
Get a label maker, or depending what you're trying to do, a roll of whiteboard tape. Also "wet erase" markers exist, that hold up a bit better to being physically wiped.
What I really want is a network switch with an extra RGB LED per port that can indicate the VLAN.
Just one RGB LED (per port) at all. Flash indicates activity, color is VLAN.
Unifi Etherlighting?
Damn, too bad I'm in Mikrotik world. Not that I would have wanted to replace my switches regardless. Of course this has me left wondering if I just got the idea from something I had seen in passing.
I'd love to retrofit some solution with an LED strip/pcb (might have to custom fab to get the spacing right) and WLED. Except there isn't much room in 1U. (syncing configs would be a problem too but I could live with that)
Although I am going to build a new router (amd64/nixos), and it occurs to me that I should make some allowances for this (even though there are many fewer ports)
I presume the more electronically inclined people could use a solder or wire jumper instead of a pencil mark.
Wow, there's something so admirable about such a well-thought design that they even considered trusting their customers to be smart enough to be able to help themselves out of being in a pickle.
I worked at 3Com in the mid 1980's. We had a smart guy (Stanford grad) working in manufacturing that was doing numerical analysis on failure rates and product returns. His analysis showed that the 3c50x PC Ethernet cards almost never failed and built a case for offering a lifetime warranty. He calculated that it would cost the company almost nothing to offer the warranty.
They were high-quality cards and the lifetime warranty proved to be very popular.
In Thailand the first TP-Link distributor offered "lifetime forever" warranty. Even if it get lightning surge you can get a replacement. For that I believe they charged twice for the product. Many people think it was a really good deal for an unknown Chinese brand at the time.
Later on TP-Link themselves added more distributor and were not happy that they cannot dictate the pricing, and also the distributor did get undercut by the newer distributors who only offer standard warranty, so they part ways.
Even now almost 10 years after that they probably will honor my TP-Link warranty but I'll get a Totolink instead
That’s interesting, thank you for they tidbit! Do you have any other stories? Personally I loved the pcmcia cards but mostly because drivers were included from 98se, so in the early 00’s my first job at a of repair shop that was an easy way to get a computer online/networked.
Outside that, I think I missed everything though I have memories of another company that merged into 3x in, xircom? Before they were acquired again.
Don’t encourage me!
I’ve already posted a handful on this item, just search for my username.
I have such fond memories of the 3c509 cards...
I wrote my first and only Ethernet driver for the 3c509. The thrill of that first ping!
There's something about interfacing to hardware that just hits different in the programming world.
is this the card that needed sometimes to be toggled to work in Linux
It is, apparently, though I don't think I've ever used it under Linux. The driver I wrote ran under pSOS, an old real-time OS we were using for a building access control product. pSOS was obsoleted by competitor VxWorks about a quarter century ago.
The fun thing about writing an Ethernet driver back then was that all of my knowledge of networking came from downloading RFCs, reading TCP/IP Illustrated and studying 3Com's 3C5x9 technical reference. Writing driver code was like solving an elaborate puzzle.
Back in the pre plug and play days, they were one of the few that would just always work.
Back in the "plug and pray" era.
The 3C509, and the following 3C905, just worked. Back before Windows 95 and plug-and-play, getting hardware to work took actual effort. But those network cards worked with pretty much everything. Then when Linux started to become a thing, once again, they just worked.
I bought a 3C509TP recently to put in my 486; the only issue I had was dirty contacts in the ISA slot. After I sorted that out, it worked perfectly.
I remember changing the jumpers on the board to allow it to enter “promiscuous mode”
I think most cards have this mode.
They don't need a jumper
That sounds like a nice scam.
Lifetime warranties are sort of a scam, mainly because they mean different things to each side of the transaction or perhaps even nothing. The customer thinks "my lifetime". The manufacturer, if honest thinks "the lifetime of us manufacturing the product", if dishonest "the lifetime of the product, when it breaks the warranty is over"
Does the manufacturer have to define what "lifetime" means? Perhaps buried in the small print somewhere.
I didn't see the thread about the lifetime warranty but for 3com stuff - means there is still firmware being released and I can call HP enterprise and get a new switch even though my current gear is old enough to legally drink. I had a bunk dell dock that kept causing arp table corruption and they were able to help me with email support in 24 hours. Pretty awesome actually.
When I was young we had a bunch of their etherstack hubs and 3c509 cards - it was fun being able to watch the lights blink showing a packet coming from one port and go cascading down the stack...until you'd have a packet storm.
Lifetime is defined in US law. Most people don't know what the law says. I don't, And I can't be bothered to look it up. But there is a definition in US law.
It isn't. "Lifetime" must be clearly defined when offering a lifetime warranty, but the law doesn't specify what it means.
It came with the card. They said “if you buy this, it’ll last forever, at our expense”.
That sounds like a stupid scam unless you’re super confident in your product.
What happens to such promises when the company is consumed by another? I assume HP is still on the line for it.
HP themselves sold Aruba networking equipment with lifetime warranties and then just decided they wouldn’t honor it anymore. So a warranty change after a company acquisition sounds even more plausible in comparison.
I guess they decided no one would take them to court, and if someone did they'd just refund the value of the item (which presumably is low these days).
Which would imply such lifetime guarantee promises are worth very little.
HP used to provide lifetime warranties of their own branded gear, so they probably weren't too concerned with it.
I have a very personal connection with 3com, though it was in the late 90s.
Ron Crane which was 3com's 4th employee and designed the 3C100, 3com's first ethernet product. Many years later he hired me for my first real job of out school and brought me to the bay area. I then got to work side by side with him and David Boggs (https://en.wikipedia.org/wiki/David_Boggs). At the time I never realized how special working with these two were...wasn't every company in the bay area like this?
I stayed in tough with Ron Crane until his death in 2017. He was both brilliant and incredibly sweet. He was so proud to take me behind the scenes at the Computer History Museum to show some work he'd been going.
I know Ron has writen about 3com...but this story captures the Ron I know: https://www.theregister.com/on-prem/2013/05/28/how-god-and-u...
"Monomaniacal perfectionism." Yep.
One day, Ron (a hardware engineer from Xerox) gave a talk about the theoretical performance of an Ethernet file transfer. He calculated the end-to-end latency and data-rate including disk latency, network latency, the number of clock cycles necessary to send and receive the data, etc.
He had (overhead projector) slides that showed all his assumptions and calculations. Finally he displayed a slide with a graph showing this theoretical maximum, the actual performance of the 3Com network stack, and (IIRC) the Novell stack. As I recall, actual performance was about an order-of-magnitude lower than the theoretical max.
His point wasn't that our software division should be able to achieve the theoretical max. It was that if we (for example) got halfway there, we would know that Novell would never ship software that was twice as fast as ours.
He was very nice and scholarly about it and did not mean offense. I and many others weren't offended. But a surprising number of people were.
I'm sorry to hear that he passed.
The Computer History Museum has a nice two-part oral history of Bob Metcalf (3Com founder and co-inventor of Ethernet) back in 2006: https://youtube.com/watch?v=zKz07DdaKzw
For those with textual proclivities: https://archive.computerhistory.org/resources/text/Oral_Hist...
Metcalfe has a reputation for being brash and outspoken and it clearly shows in the interview. I think it served him well in business and later in life as a pundit. I remember looking forward to his columns in InfoWorld, though I didn't necessarily give much stock to his predictions.
He talks a little bit about the downfall of 3Com in the interview, though he begs-off of talking about things that happened there after he left.
I very much enjoyed the trivia and minutiae about the creation of Ethernet and the Digital / Intel / Xerox relationship that ultimately resulted in it being standardized.
Basically, it was a long listen but there was some good stuff in there.
It was great to work with Metcalfe. I have two funny stories to share.
I was on the Macintosh network os client team. For whatever reason (maybe he was annoyed that Apple copied so much from Xerox) Metcalfe didn't like Macs. In this era Macs used Apple's proprietary AppleTalk/LocalTalk network. We did a demo for 3Com execs using a large external monitor connected via a single black coax cable that looked just like thin-ethernet. He got excited for a moment and came right over to look at the Mac and was very disappointed to see the cable attached to a display! (3Com was working on Ethernet cards for Mac at the time but they weren't ready at the time of this demo.)
For those too young to remember: there was an SF Bay Area baseball park known as Candlestick Park that was briefly branded "3Com Park" -- to the great annoyance of SF Giants fans. During one interview he as asked if in the early days of 3Com he would have done anything differently. He said, "I would have named the company 'Candlestick'!"
Ha! He makes the Candlestick remark in this interview too.
> though I didn't necessarily give much stock to his predictions.
Every year that I worked at 3Com, we proclaimed it was "the Year of the Network!" It became a running industry joke.
But of course, Metcalfe's Law contains some fundamental truths: https://en.wikipedia.org/wiki/Metcalfe's_law
I remember those days, with Computer Currents' "Year of the LAN" parody predictions repeated year after year. Macs with PhoneNET definitely made it easier and cheaper in the beginning.
It's like proclaiming The Year of Electricity. You can proclaim it once, say, in 1900, and it just would never end since then.
Proclaim the Year of the Network some time in 1993 or 1994, and you're golden.
They first did it in 1986 -- and it was too early!
(It was the 1980's version of the Linux Desktop joke)
And the products got shitty. They sold a managed switch that couldn't save its configuration
Do you remember which switch models it did happen to you?
I remember that SuperStack 2700 ethernet switches with ATM LEC (lan emulation client) card it did happen under prolonged heavy load heat accumulated with then caused sudden reboots. Occasionally these events switch was as factory reset was done, none of configuration made was left.
First with ATM LANE with ethernet STP (802.3D) was flimsy and buggy, so it took a while when that was fixed without STP spurious loops broadcast storms did quite easily overload edge switches and cause those sudden reboots.
We had to keep minimal ethernet management part configuration printed next to switches so that first entering primary network settings we could TFTP-load production config and get switch back to production quicker. But long without OOB management (serial port & remote telnet) it was always was hurry to get on site fixing locally these events.
I wasn't earning those occasions much after we got most often issues caused switches replaced with another other brand with did not suffer these problems.
By design?
That is a remarkable decision.
> They sold a managed switch that couldn't save its configuration
I don't hate that idea. I wish I could set up our Cisco switches to do that, so on startup they pull down their config via ZTP and never ever ever save local changes.
That'll stop people logging in and making local changes and expecting everyone else to somehow just know about their tampering.
>That'll stop people logging in and making local changes and expecting everyone else to somehow just know about their tampering.
How about changing "enable secret" password and not sharing it with those who you do not approve making changes?
Or even setting up TACACS+ or RADIUS service and getting SSO to network gear that way. TACACS+ even free versions by default as an AAA that last A accounting makes you able to log all logins and commands being executecuted IOS cli.
Also, you can set up a management server (station) which collects all configs, checks in a db or VCS (RCS, CSV, git) which you can then build on changes monitoring, revert back configs as far as you have saved config versions. Like I had while working over 20 years backlog about half thousand switchers and other gear. RANCiD can do it for cheap, if you don't like to buy commercial solution which also can do similar things. Oh, OpenNMS and FreeNMS probably can do this also if you prefer getting those. I didn't use those as we built from mid 90's own systems and didn't bother changing at least when I was still working.
If you still need to allow other users login let other commands that enable level requires, you can use a "parser view group-name" and pick what additional commands you let them also access from exec enable level (15) without letting then become full enable 15 level users.
ps. Cisco IOS has long had service which autoloads config from TFTP server, but that isn't perhaps what you are looking for. It's mostly intended getting initial config uploaded to system enabling remote management over telnet and further making proper config automation easier.
> How about changing "enable secret" password and not sharing it with those who you do not approve making changes?
Ideally the management level wouldn't be on the tools, yes.
> Or even setting up TACACS+ or RADIUS service and getting SSO to network gear that way. TACACS+ even free versions by default as an AAA that last A accounting makes you able to log all logins and commands being executecuted IOS cli.
Sure would be nice to keep the management level out of RBAC
> Also, you can set up a management server (station) which collects all configs, checks in a db or VCS (RCS, CSV, git) which you can then build on changes monitoring, revert back configs as far as you have saved config versions
Yes, that's how I know who to shout at.
> later in life as a pundit
An amusing example of his punditry:
"I predict the Internet, which only just recently got this section here in InfoWorld, will soon go spectacularly supernova and in 1996 catastrophically collapse."
Keep in mind he co-invented Ethernet, so he knew what he was talking about!
Here's hoping we'll be able to look back with equal amusement at the AI doom predictions.
https://www.elon.edu/u/imagining/expert_predictions/from-the...
Prior to the time period covered in the article, 3Com had a major disaster in Software. We were losing badly in competition with Novell in the Network Operating System category (though I and others would argue we had a technically superior product) and decided that we would partner with Microsoft and overpower Novell that way. Big mistake.
I was only a consultant at 3Com during this time, so I may have some of the details wrong, but 3Com essentially bet the success of its software division on the partnership. We built a product called 3+Open that was based on OS/2 Lan Manager. The product flopped. But unfortunately 3Com had agreed to multi-million-dollar, fixed annual license payments to Microsoft for OS/2. (Which may have been the right choice had the product been successful.)
To get out of the license payments, 3Com gave all its LAN software IP to Microsoft and exited that business. Microsoft used that technology in Windows NT Lan Manager and essentially destroyed Novell.
If you want details, the Bill Gates Biography "Hard Drive: Bill Gates and the Making of the Microsoft Empire" tells the story of that failed partnership.
I worked for a while doing pre sales support for a company that sold 3Com products. 3+Share and 3+Open were pretty solid but mostly people just bought Netware. The XNS network protocol couldn’t be routed, only bridged, our biggest customer had 3Station disk less PCs and a huge number of 3Servers on a huge national network that gave them headaches managing due to broadcast traffic travelling over slow links. It was a fascinating problem to have in the era before properly smart gear.
XNS protocol supported routing via IDP (Internet Datagram Protocol) though there were issues and limitations especially when compared to IP.
I’m not doubting that you couldn’t route XNS with your implementation/configuration.
“broadcast traffic over slow links” brings back memories.
Microsoft LAN Manager was a flop, too. Windows NT was entirely new product. No such thing as NT LAN Manager existed.
Source: me, LAN guy since 1990 and NT admin since 1993.
Yeah, I probably got the product name wrong.
I don’t think NT was entirely new at the source-code level, though.
Update: The product was called Windows NT Advanced Server:
https://archive.org/details/ms-winnt31-as
Which in some beta versions was called "LAN Manager for Microsoft Windows NT", so they intended to keep the branding at one point. This carried through to release in a few ways, there's still a library called ntlanman.dll in system32, the legacy authentication scheme you should never use on windows is NTML authentication, and on NT 3.x/4.x the server SQU wallpaper was lanmannt.bmp / lanma256.bmp.
Not sure if there's much direct carry over code in NT 3.1's networking code from OS/2 though.
> story of that failed partnership
For Microsoft it sounds like it "failed successfully".
Never heard about 3Com work with MS before... Just WOW ! Another company cooperation with MS failed and then MS use wreckage to big gain.
MS realy is the trojan horse. Because probabilities multiply into less probable outcomes but result is again and again the same.
>Another company cooperation with MS failed and then MS use wreckage to big gain.
Indeed deal as good as the DEC one.
Recently saw 2 different network closets one filled with old 3com and HPE switches and the other one was HPE and Aruba. Basically the same switches, basic 24 port 1GbE. Just feeding a bunch of cameras with poe. Surprised the resilience of 3com hardware. Surprised how little changed in-between these generations.
First time i saw 3com's they were already ancient. One of the switches had a coax cable that connected to a hut that was too far away for Ethernet, wifi was not permitted, no one wanted to pull an optic, so we basically replaced everything around that hut an that switch with Mikrotik and just let it to chug along way past it's expiration date.
I worked at Convergent Technologies at the time, and our CEO went on a vacation cruise and met Metcalf. They decided over drinks with little umbrellas in them, to merge our companies. We were the largest single customer of Intel processors at the time(the PC market was very fragmented).
But the stocks jiggled and the buyout plan was postponed when valuations became uneven, and it never happened.
The 3c509 was the card we slapped into computers when the cheap NE2000 clones wouldn't work. Always did.
Years later I had a EtherLink III MCA 3c529 at home. I think it was one of two Microchannel cards they offered. Corporate sites usually had the IBM Lan Adapter/A since they were typically True Blue.
I also had a Office Connect ISDN LAN Modem 3c891 during the year I had working ISDN through Bell South. It supported channel bonding so I could get 128kbps, and it would automatically dial my ISP [0] whenever a machine on the local network wanted to go on the internet. Truly useful.
[0] Uzi Nissan, of nissan-dot-com fame. RIP.
Was just thinking about nissan-dot-com the other day. Sad to see that man has passed but overall pleased it’s still his.
network cards mostly died when it got integrated into the motherboard - same with soundcards and even graphic cards unless you wanted pro quality sound, extra networking features or a gamer. For everybody else, the basic integrated options were enough. I grew up in that era and not once did I look up what the 3com card offered over another networking card other than 10/100. Just wanted it to work. I'm sure some people are more discerning but not much.
There were some “interesting” issues with a lot of the cheap hardware for a long time though. A classic example was the WinModem. A fast modem for cheap? Sure! But it was cheap because it outsourced a lot of its processing to the CPU, so if your system was loaded up (in the day when CPUs were almost all single-core) meant there wasn’t enough left over to process inbound data in realtime which meant dropped packets and retries and horrendous lag and throughput.
Standard wisdom was to use an Intel EtherExpress Pro card for a long time after onboard Ethernet was available. Lots of server hardware had unused builtin NICs until they got decent.
The separate cards may have died, but the same companies, or different ones, are making the chips that go on the motherboard. I think mine has a 2.5G Marvell and a 10G Intel chip.
>not once did I look up what the 3com card offered over
https://news.ycombinator.com/item?id=42746741
tldr:
"The RealTek 8139 PCI NIC redefines the meaning of 'low end.' This is probably the worst PCI ethernet controller ever made...8139 supports bus-master DMA, but it has a terrible interface that nullifies any performance gains that bus-master DMA usually offers.
For transmission, the chip offers a series of four TX descriptor registers. Each transmit frame must be in a contiguous buffer, aligned on a longword (32-bit) boundary. This means we almost always have to do mbuf copies in order to transmit a frame, except in the unlikely case where a) the packet fits into a single mbuf, and b) the packet is 32-bit aligned within the mbuf's data area. The presence of only four descriptor registers means that we can never have more than four packets queued for transmission at any one time.
Reception is not much better. The driver has to allocate a single large buffer area (up to 64K in size) into which the chip will DMA received frames. Because we don't know where within this region received packets will begin or end, we have no choice but to copy data from the buffer area into mbufs in order to pass the packets up to the higher protocol levels.
It's impossible given this rotten design to really achieve decent performance at 100Mbps, unless you happen to have a 400Mhz PII or some equally overmuscled CPU to drive it."
tldr tldr: it didnt matter past 2001 since slowest low end CPUs were cheap enough to nullify all the potential gains from smart good 100Mbit NICs
Sounds like it could have been fine with a different software stack, which did not exist.
No amount of software would have helped. It had limitations as to where it could fetch data from and where it put received packets, both requiring copying all network data around. Plus shallow TX queue resulting in higher interrupt load. Afaik 3Coms had proper descriptor rings DMA Master engine allowing zero copy and deep queues.
You couldnt saturate 100Mbit link with weak CPU on Realtek while 3Com did all the data shuffling for free. It didnt matter when browsing internet on contemporary slow links, and stopped being noticeable when copying data over Lan on >400MHz CPUs.
Did anything stop you from processing the packets in the place they were received and putting the data you want to send in the place it could be transmitted from?
They made good stuff. Then the cheap stuff got good enough for 90 percent of the use cases, then they bought into another sector where exactly the same thing happened again, except it was also a completely dead end. I don’t think any peripheral-card-only makers survived except Nvidia
I installed a bunch of 3Server network servers in the late 80s-mid 90s. They were pretty darned solid. Tech support was an actual human, I knew his name (though have forgotten it now), and we actually went out to lunch one time, as my company was in Mt View and they were in Santa Clara, I think.
Well there’s a name I haven’t heard in a long long time
I always figured 3Com just faded away when PC motherboards came with Ethernet built in.
I used to have a 905 on my router's external interface, until one week I had to do a modem swap. A couple days after the guy left, I went back in to do some cable cleanup... and when I plugged the cable back in, the surface-mounted jack sheared right off the pcb. While I could have soldered it back on, I'm surprised most cards didn't meet the same fate early on.
(2022) Yeah.. I know, it says 2026 at the top, but it was posted here in 2025, and you can see it's older. No idea why the author keeps ever-greening (the create date should at least reflect the original post?)
(Small text changes, and a CTA at the bottom)
https://web.archive.org/web/20220302141059/https://dfarq.hom...
SEO issue, changing the byline dates makes crawlers think it's new/fresh. Google prioritizes "fresh" content.
Those of us who were poorer had to settle for a cheap clone of the NE2000; the 3c509's were too expensive.
"never diversified" is a funny epitaph for a company that at various points owned Palm, US Robotics, half a Huawei JV and an IPS vendor.
Yes, bought TippingPoint[1], one of the first NGIPS makers. That were one of first very capable L2 IPS to market. I did have opportunity to test one for a while, but did not get funding for it approved.
3Com was also one of the early ATM gear vendors with CELLplex 7000 ATM switch and SuperStack 2700 ethernet switches with LEC. I would say mostly[2] reliable and working gear few years before Cisco LS1010's ATM switches, Catalyst 5000 and first 3000 series ethernet switches after acquiring Kalpana.
1) https://en.wikipedia.org/wiki/TippingPoint
2) SuperStack switches had bad habit suddenly reboot under prolong3ed heavy load and perhaps heat problems too, which then also occasionally did factory reset and lose configuration. And the menu based configuration with binary only tftp upload/download configs when better alternatives came available these were replaced any location where that ever happened causing issues.
Quite a long winded way of saying: bits of HP(E).
I can't quite remember if the 3c509, 595 or 905 was the fastest and best. I think 905. I created a boot disc for these and a few other NICs that featured on Novell's Cool Solutions.
Nowadays I have QSFP28+ to fiddle with.
I remember also original EtherLink 3C501 and 3C505 with 3+Share yay!
Both were quite exiting and beautiful looking new fancy cards early -85 when I first had them on my desk. But in reality 501 was a dumb as a dodo and 505 with its dedicated 80186 (co)processor for 3+Share fileserver intended use not much better.
Here are links to pictures of those wonders we got -80 and cursed them later :)
I bought a 3c595 FastEtherlink card from CompUSA when I was in college because my dorm had ethernet and I was running FreeBSD and I thought tech specs were the be-all end-all.
It was fast, but not noticeably faster than any other cards my friends had. Except I spent way too much on mine.
3c509 was 10base-T ISA, so it's the slowest. 3c595 and 3c905 are 10/100 PCI cards, but the 905 was newer and more efficient (better bus mastering, more CPU offload)
This is a very, very common story for many tech corporations. They find a place then fail to adapt. Adaptec (ha ha) is another example that could not adapt.
This inability to adapt is inherent in the corporate business model. Look at Apple. They have a chicken that lays golden eggs. Then the chicken dies. Oops. So they invest in speculative funding of massive proportions. Just ask AI and it will tell you.
Maybe there needs to be some other form of business that can do better?
Didnt Adaptec invest heavily in Fibre Channel? From what I remember from Computer History Museum CEO interview they went all in on FC and didnt see a point investing in InfinityBand as they saw it as also ran.
Article is a bit wrong, it's HPE that owns it now, not HP. Comware switches are still actively developed and sold. It's very solid, deployed in places you wouldn't expect where it just runs for years without issue.
What happened to them? They never really diversified out of network cards, their cards were very very good, but for most people paying 3x as much wasn't worth it.
They split off their Palm (PalmPilot) division in 2000 to focus on Network equipment. The the .com bust happened.
Palm/Handspring/Treo could have been the iPhone, if they'd been better managed.
Could have, should have, would have…
iPhone wasn't just a form factor, the software traces back to NeXTSTEP 1989.
Fact of the matter is nothing could have been the iPhone, everything sucked in 2007.
Apple had the wind in its back with the iPod and Mac OS X Tiger.
The Intel transition was complete.
Apple was firing on all cylinders, what a time to be in your early 20s.
Even Android did a 180 once the writing was on the wall, it originally resembled BlackBerry OS.
Apple was also powerful enough (and Cingular at the time, weak enough) that they could dictate the terms and do whatever they wanted. Anyone else would have had their vision ruined by the cellular carriers.
I had a Treo fairly early on. Leave aside the physical keyboard. The iPhone had a whole app and media infrastructure built around it. Could Palm have gone there? Possibly. But that would have been a big step from where they were in the mid/late 2000s.
Android was originally designed for digital cameras, with tactile interfaces.
The camera bit is true, but I'm not so sure about tactile interfaces?
Palm had apps from the start.
I loved my Treo, and in some ways it bettered the iPhone form factor, but the software wasn't comparable.
but the windows phone OS was better, and the hardware very good. They just couldn't get their shit together with the app ecosystem; even paying us devs huge premiums to write apps wasn't enough.
Nope. Innovation doesn't happen that way. Once they took on the keyboard path, it's nearly impossible to innovate into something that's not. Even if a team is innovative enough to come up with it, top level will kill it off
3com was the very first investment I ever bought…I bought my shares on Thursday October 15th, 1987….those that are old enough to remember will know how that turned out.
I can maybe do you one better. I think my first investment was 140 shares of LNUX at F&F IPO pricing of $30/share, purchased Dec, 1999. It opened at $299/share the day of the IPO. Should've sold that day. Eventually sold it all in 2004 for $2.80/share.
https://en.wikipedia.org/wiki/Geeknet
A fun article but it is incomplete. It misses out on the company's height.
In the early days it has its own server OS, 3Com 3+Share. It was a multitasking file server OS and it was based on MS-DOS. Yes really. An internally multitasking server stack on DOS. I became expert in it in my first job and my 3rd job leant heavily on that and my skills.
It used the MS LAN Manager stack over NetBEUI.
It inherited DOS limitations like 32MB partitions. On PC servers everything had to for into 640kB of RAM and natively it could not use XMS, only EMS.
3Com offered a range of dedicated network file server boxes, the 3Server. The first was an 80186 and later the 3Server3 with a 386.
Then it got big into OS/2 and released 3Com 3+Open. The multitasking stuff obviated by OS/2 1.0's own built in multitasking but now it was multiprotocol and the server could run other stuff too.
The 3Server had no display, no port for one, no keyboard connector. It could only be administered over the network. It did have a little 1-line LCD display for showing a status message, accessible via a DOS device driver for showing custom messages. A physical rotating dial let you choose a station number at hardware level.
Some pics here, but not the models I knew:
https://www.bcsatellite.net/blog/pats-memoirs-design-time/
Because there was no display, that part of the 80186 memory map could be used for RAM, so a 3Server could give 800-900kB of RAM to DOS. A selling point: it could have 25% more memory than a PC and was arguably easier to configure.
The 3Server3 had for the 1980s very high memory limits, but we didn't sell many.
3Com's DOS apps had their own pre-CUA text-mode windowing and menuing system, 3+Menus, with a consistent UI and navigation keys, before Windows 2.0. Fill in fields in any order, Enter to accept, Ctrl+Enter to accept the whole dialog box, close it (with an animated zoom-out effect), and return to the previous level. This applied to all 3Com software and meant very fast navigation. Last seen in the DOS setup tool for the 3C589 ISA NIC.
As well as file and print, it bundled an email server client, which almost none of my clients ever used, and it also could act as a network gateway, with tools for WAN linking of networks into a big MAN, which I also never used.
I remember putting 3+Share on an IBM PS/2 Model 80 with the biggest hard disk I'd ever seen at that time: 330MB, on an ESDI interface. I had to partition it as drives C:, D:, E:, F:, G:, H:, I:, J:, K: and L:. All 32MB except for an 11MB L: drive.
NOS on C:, main network app on the next letter, data on the next letter, home directories on the next, print spools on the next... it tool some ingenuity to find ways to use them all on DOS 3.3.
There was only one supported big-partition tool, VFeature Deluxe from Golden Bow Systems, and none of my clients were willing to pay for it. GBS did various DOS addons: a software disk cache, VCache; a defragmenter, VOpt; and VFeature, for disk partitions over 32MB.
https://golden-bow-systems.software.informer.com/
I even tried installing it on Compaq DOS 3.31 which had big FAT16 partition support using the same system that MS-DOS 4 later adopted, but 3+Share could not see the big drives.
wow ! that sounds like lots of fun to be able to mess around with systems like that at their hey day
Same thing thst happened to Digital Equipment Corporation - acquired by HP's bad clueless 1990s/2000s CEOs.
Digital was acquired by Compaq before it got purchased by HP.
I knew someone was going to point this out, and I still explicitly didn't mention it. Dance, puppets.
Had my first 3c509TX (I think that eas the suffix) sometime around ‘98 or ‘99. It was a luteral fortune back then for me.
But it worked until the other components in PC stopped working first.
Fond memories.
Hp bought them ;-)
btw what happened to US Robotics out of curiosity
(very expensive external 56K modem sitting in my junk pile over here)
* https://en.wikipedia.org/wiki/USRobotics
ah equity acquisition, the kiss of death
The article says:
>The first thing that happened was 3Com walked away from its enterprise networking equipment business. Admittedly Cisco was hard to compete with. But 3Com was the only brand succeeding at doing it.
I can't really agree with that because 3com never had a serious router product line. They had LAN switching equipment and NICs. There was nothing like a whole product line of routers that could take a full BGP table in the 1999-2000-2001 era, and certainly nothing like medium to large carrier routers. I'm talking about around the same era when the Juniper M40 was released to compete with Cisco in that market. Or a Cisco 12008. Anything broadly along those lines. 3Com was far behind in advanced IP routing and things that would serve as like a metro aggregation router or core router.
I don't know, we were eager to use 3Com products over Cisco whenever it was an option. Cisco was a bit user-unfriendly, not to mention expensive.
But for what application? LAN switch? Endpoint CPE router equivalent to a 2611? I never saw a 3com router that could take a full table.
A few lessons from my early IT life, all by way of 3Com...
A strange OSPF problem with 3com NetBuilder routers of which nobody could get to the bottom of. Based in a team that maintained the 3Com relationship with this arm of the (UK) civil service, who solely used 3Com networking products (LAN & WAN), we duly rang our contact and described the problem (and attempted fixes)... "You should change the PSU(s)" he said (eh? it's an OSPF problem?!?!?). Apparently they had some sort of (RF?) issue(s) with certain NetBuilder PSUs that had been known (note: undocumented, at least not publicly) to cause "strange" issues (it did resolve the problem).
Lesson 1: Sometimes it's the thing that wouldn't even occur to you that are the root cause.
Having ordered lots (>£1mn) of 3Com enterprise networking equipment (from LAN switches to their CoreBuiler 9000 chassis), and stuck it all in a warehouse, in preparation for a country-wide network upgrade - 3Com shortly thereafter (and seemingly out of the blue) announced their complete withdrawal from the enterprise network market. 3Com's messaging, as I recall, was pretty much - once the existing stock is gone there will be no more, and hardware support would last 2 years and after that - tough! Needless to say this caused quite some consternation with civil service execs, of which I assume invited 3com to a meeting (without biscuits) to discuss the situation. 3Com eventually partially walked back on said withdrawal (continuing to manufacture/sell smaller variants - and card/PSU/etc compatible - of the CoreBuilder 9000) although this wasn't enough to un-sour the relationship.
Lesson 2: Maybe this isn't the best way to treat your customers, and no matter how large the vendor (or how much you spend with them) - this may happen to you.
In one of 3Com's offices (Partington, a delightful place) - if you parked near the edge of the, fenced off, office car park - you may return to your car to find a brick embedded in it as a gift from the one of the locals. I assume much merriment was had by the cheeky scamps when tossing bricks over the fence at 3Com employee's cars.
Lesson 3: Get to this 3Com office early and park in the center of the car park.
Sat with a (TUPEd ex-civil servant, and hence un-sackable) senior network specialist who was updating the network monitoring for all devices in a specific region. This involved refreshing (via SNMP) the interface list of all the routers/switches/etc. During this process, which was basically click refresh - move onto the next device - and then revisit the original device after a while to confirm it had completed, one of the edge routers dropped offline (and didn't return). No problem - traffic would route (and bridge! IPX FTW!) via the backup router/link. The senior network specialist was quite methodical in his polling, moving from one (now offline) edge and around the ring of devices before eventually arriving at the final (note: backup - and now only - route) edge router (NetBuilder). Prior to reaching the final router it had dawned on my little (junior, green) brain that maybe he had triggered a (documented) NetBuilder bug that would cause the router to hang - needing a physical reboot - if SNMP walking a device with an FDDI card (which these routers had). Having voiced my theory, and in doing so suggesting that polling the final router may not be the wisest course of action, he confirmed that it was his job to poll the routers and that is what he was damn well going to do. The backup router dropped offline. 100s/1000s of civil servants now unable to do their job (that required mainframe access that resided beyond the edge routers) and needing physical intervention (i.e. ~1 hour) to resolve. Based on previous observations, I assume productivity for those users dropped by a single-digit percentage within that dark hour.
Lesson 4 (one of many such lessons reinforcing this view): 3Com's enterprise networking software sure has some strange bugs. Also, welcome to the public sector (or public sector adjacent).
Given the various bugs observed during my time in the team - I didn't hold the quality of 3Com's enterprise networking products' software in very high regard. Such low opinions would only be surpassed in later years by Fortinet's security/networking products' software releases.
I am sure that the built-in Intel Ethernet ports on most motherboards have made Network Cards obsolete at least for the Home Market. When I got a 2Wire DSL modem, they threw in an Intel Ethernet card with it for free. Now I use WIFI 6 on my Wifi router and PC, and it is fast enough for me.
I remember when USRobotics was giving away free 56Kbps modems to BBSes if the SYSOP listed that they had a USRobotics modem and kept the BBS up for 24 hours a day.
> I am sure that the built-in Intel Ethernet ports on most motherboards have made Network Cards obsolete at least for the Home Market.
Someone is still making a NIC chip, they're just selling it to the motherboard OEM, instead of selling an add-in board to the customer (or selling a chip to the add-in maker). In a way, this is more sales, because desktops tend to come with a wired nic and a wireless nic, and so do good laptops.
You are understating how integrated this stuff is. If you make an Intel client motherboard, all you need to add for the network is the phy, jack and some passives.
I haven't really been following intel boards. AM4/AM5 chipsets (and cpus) don't have integrated MACs. I didn't realize intel was putting them in for client parts (Xeon-D had them, but Xeon-D is weird :p)
https://en.wikipedia.org/wiki/CNVi