Slashdot Mirror
Open-Sourcing Discontinued Hardware
LinuxWhore asks: "I work for a company that recently accquired two 3Com/USR TOTALswitch units. It seemed as though we had I nice product by the price that they were going for online ($1500-$3000). However, further research had revealed to me that 3Com had decided to discontinue all work on the line shortly after their merger. All updates to the product have thus ceased. Now I am left in a situation where the product has little documentation and no chance of future security/bugfixes. If companies like 3Com were petitioned to release the souce and hardware specs to their dicontinued products, how much interest would there be in the community to write updates to code for these types of products so that they remain useful, instead of becoming a $3000 doorstop?" It's a good idea. Convincing the hardware makers will be the difficult part.
If you just bought one recently for $1500, you got the shaft (tm). We picked one of these up on auction 2 years ago for $300. 3com is no longer supporting them, but that doesn't really matter. The latest firmware revision works fine. The only problems are that the 100bt cards don't support full duplex, and there's a backdoor that you can get into through telnet (see my bugtraq post from a few years back). This is easily solved by either not setting a default gateway, or firewalling access to the telnet and www ports from the internet (which you should do anyways with this kind of equipment). Our Totalswitch is still working great. We never needed any support from 3com. Most of the information that you need can be found at totalservice.usr.com (you'll need the serial # of the switch to get a login/password). Or on-line. Adam Maloney Systems Administrator Internet Exposure, Inc.
IMHO, if you can convince a hardware manufacturer that there's gold in them thar doorstops, they won't be able to open-source fast enough. In the end, the "new" products aren't what's important. What's important is that the CEO's piggy bank gets one step closer to gravitational collapse.
How to do this with Open Source? That depends. Research Machines, in England, are a good example. They've made some stunning, innovative machines and components, in the past, which they have subsequently thrown away.
Open Sourcing the drivers, at the retirement stage, could help clear the surplus stocks. After all, they can then flog off the stuff they'll never shift to anyone else to student-types and Linux companies.
What about more internal stuff? BIOSes, etc? Undocumented calls to those fancy non-standard chips? Again, there are going to be surplus in stock that can't be shifted to retailers any more. They'll NEVER get rid of them that way. But once the attention's fading on the new releases, again, they could Open Source virtually everything on the older models. Empty those warehouses, at no expense to the company.
IMHO, free money and new customers are gold to any company. If it's done in a way that doesn't hit sales of newer products, because it's aimed at people who won't buy the new products anyway (because the drivers don't exist for Linux, *BSD or BeOS), you've wheelbarrows of green bills coming in, none going out and all that expensive storage lots can be used for stocking something profitable on the mainstream market.
Money Talks. It even sings and dances a little, if you let it.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
Why is lead solder still so popular? 60/40 lead solder is cheap. There's also another evil to lead solder: it tends to crack over time under heat stessed components. This is why over 50% of televisions fail. It makes sense for companies to sell a product that works very good at first and start to have intermittent problems after a predictable 4 years. 60/40 solder is the perfect time bomb for this.
Lead solder is the most accurate time bomb for those 3 year warranties.
Ever shake a malfunctioning television that starts to act flaky after a few years you bought it? Resoldering the problem areas almost always fixes these problems. I used to fix televisions and big screens for $150 up to $450 each when it was very profitable several years ago. To keep the recall rate low, I used silver solder. Of course, I checked if the overheated capacitors were out of tolerance which can also aggrevate the hot spots.
The problem isn't just hardware. Books, code, and all sorts of other things go obsolete, discontinued, or out-of-print. Have you ever tried to get a copy of a long-since out-of-print book -- you'll pay thirty times the sticker price. Out of print CDs are less expensive but still far pricier than current ones.
I have often wondered if there might be some way -- social, economical, or legislative -- to force things into the public domain once the owner isn't using them any more. Why on earth should it be illegal for me to digitally distribute a book that's been out of print since 1940? After all, one can hardly claim that I'm costing the publisher anything.
Deliberate obsolescence of a product -- particularly if that product is copyrighted or patented -- needs somehow to be turned into public-ownership of that product.
Perhaps copyright should expire at author's-life-plus-n (setting aside the whole "what should n be" debate for now) or the point at which the publisher is no longer willing to publish, whichever comes first. Something similar for patents might prevent certain offensive legal strategies involving patents.
I'm not one of these zealots who thinks that we should all be downloading copyrighted this-and-that for free without legal reprecussions, but when something is out of print, no longer manufactured, or a patent sitting idle in someone's files, it really should be fair game.
--G
Many new hardware designs use FPGAs for substantial portions of the board logic. The source code, in the form of a schematic or other high-level representation, is compiled into a binary file that is loaded into the FPGA when the board is initialized, either from a PROM or by the system's CPU.
Mea navis aericumbens anguillis abundat
I have a five year old NEC desktop. I figured it would be stupid to waste this old system so I stuck Linux on it and made it a dedicated file server. Then I got my new hard drive and tried to put my old one in the file server to give myself a bit more space. The NEC wouldn't accept the second hard drive. Calling NEC to ask for hardware specs or something that might help me (maybe a BIOS update) they just told me to go shove my hard drive up my ass. I don't see why a company can't provide at least information on systems that are obsolete. Why hold the specs to obsolete and non-competitive product so close? Video cards are another thing, why don't the chip and card makers release free specs for their older cards that they don't even produce. Even under and agreement that says you'll only use the specs to write open drivers or some such would be nice. Sheesh. fnord.
I'm a loner Dottie, a Rebel.
Good point. However my counter point is that some of us do not buy hardware from companies that do that. In fact, I lean heavy in the buying department to make sure we favor companies that have some form of open source policy that I and most ./ readers would like.
As the market gets tighter, companies soon find themselves without customers who are sick of this tatic.
Linux O Muerte!
3Com isn't offering support on these switches as it is. A company I worked for picked up a couple for a song (onsale was unloading them). We had our standard 90 days of support in which we burned the hell out of them and shook out most of the bugs, but after that we are on our own.
As for who is interested? I am. If I could get the source to the firmware, I would love to fix some of the annoying things this box does. But without the source, I have nowhere to begin.
Open source is much bigger than Linux. You are complete right that only those in my situation would want to work on the project. But I'm sure there is at least one other person besides me. Every added person increases the number of improvements and testing over what I could do by myself. None of us are interested in selling our improvements to the switch, we just want the farging switch to not suck so we can get the rest of work done.
Funny thing about proprietary hardware - some companies never bother releasing specs after being dissolved or swallowed, and customers wind up getting the shaft. I once bought a DSP modem that was "upgradeable" with firmware upgrades.
Pift! Gone was Cardinal, and with them those potential upgrades.
I found that out about my USR TotalSwitch a year ago. Really disappointing that though the hardware supports full-duplex they never released the firmware to support it. Ended up selling my TotalSwitch on ebay and buying a couple of Kingston 10/100Mbps full-duplex non-managed switches for what I got out of it.
That's too extreme: company officers are given pretty much complete freedom to decide how to pursue shareholders interest: if they think that the goodwill created by opening specs is a good investment, that's their call. Also, in the UK at least, it isn't illegal not to pursue shareholder profit. Instead shareholders have the right to kick out executives they don't think are doing well.
The capital investment in hardware development is far too great to go and reveal all the inner workings of older models when there is still competition against your newer ones. Even if a product line has been discontinued, individual components or concepts from that line may be reused in newer product lines. Here's a hypothetical situation:
Say for example your company produced the Foonmatix 1/10, a 100 port 10-baseT Ethernet switch. Buried deep within the switch is a rather elegant circuit, the Foontek 3842 that knows how to efficiently redirect a packet to, for example, port 32 if port 33 has a collision on it.
Five years later, you've discontinued the entire 10-baseT line. Now you're producing the Foonmatix 4/100, a 400 port 100-baseT switch, which is selling like hotcakes, because it has an incredible new chip buried in there that does predictive packet routing so efficiently that you can handle nearly 50% more traffic than equivalently priced switches. So why aren't you open-sourcing the Foonmatix 1/10?
Turns out that the only way to do predictive packet routing efficiently is if you have a good way of handling wrongly predicted packets. For everybody else in the business, a misrouted packet is a nightmare that grinds the entire switch to a halt for nearly 15ms, an eternity on a heavily loaded network and one that until now has been handled by putting a buffer on every port and redesigning the switch so that it can handle the datastorm that happens when all the buffers are freed at once. An expensive, complex mess. Enter the Foontek 3842's 100mbit descendent, the 3842A. When hooked up to the predictive routing circuit (PRC), it (a) quickly sends the bogus packet where it is supposed to go and (b) tells the PRC that it's misrouting packets from port X. Slicker than goose poop, and it wouldn't work without the Foontek 3842A.
So: when you open-source the 1/10 switch, you end up open-sourcing, or at least describing, the Foontek 3842, which has since become a critical part of the switches you've bet your company's future on.
It's too easy to give away trade secrets indirectly. In a field as competitive as computer and networking hardware, the risk of giving a competitor the edge by effectively giving them all your old ideas is just too great. What if the design you toss out the window is the magic bullet your competitor has been looking for?
Personally, I would rather see this happen on the software end of things. I can think of a half dozen old software packages that were excellent in their old environment and could become portable wonders if open-sourced. Borland's Sprint editor and Microsoft Word for DOS version 5.0, to name two. I have my doubts as to whether either of these products contain super secret algorithms that matter now, eight to ten years after they were dropped.
--
This is not my sandwich.
Man I used to LOVE pouring over the schematics and programming info that hardware used to come with so long ago...and then, ever so slowly...they just...stopped.
Goddamn shame...I certainly hope the Linux community doesn't go that way...all source and fun...and then, slowly, more and more closed...
:(
-- www.bteg.com | bleh.n3.net | hac47.dhs.org
However, since devices such as switches [as was the piece of hardware originally mentioned] and routers are normally based on highly custom architechtures then of course you'd require a custom compiler for each piece of hardware to take whatever source language you use and turn it into the appropriate binary language for the device.
The compiler to do this could well be the same piece of software being used on said companies newer products so they could well be unwilling to release it meaning the first task of anyone playing with the hardware would probably be to produce the compiler and given the pace of development in the network kit world by the time that has happened the kit would probably be totally obsolete.
J
I am not a Frog. I am a Free Womble!
In order for 3Com to make this decision, it would have to be profitable. Currently, 3Com makes money selling new hardware, and loses money supporting hardware no longer on the market. Additionally, if they support hardware not on the market, they create an incentive for people to not buy new hardware. As a result, releasing specifications will decrease their value.
But we want open hardware, right? Well, the solution is to propose a scenario by which it is profitable for 3Com to open their hardware.
The solution is for 3Com to announce that ALL new products will be supported for X years, after which their specifications and code will be released into the public domain (or GPLed if they are worried that Cisco would grab them). As a result, new hardware buyers would know that their products would work for AT LEAST X years and perhaps beyond that.
With Cisco or other competing gear, customers have no idea how long they will work. As a result, customers have an incentive to purchase 3Com gear because it has a longer lifespan. This should make 3Com gear more valuable, and theirfore be in 3com's interests.
Now as to existing hardware? Perhaps 3Com could make this somewhat retroactive. Why would they do that? Well, if groups of developers support their old gear, than 3Com's claims of longevity are more likely to succeed. They have a reality.
Also, this community, as other posters mention, consists of large numbers of technical people involved in purchasing decisions. As a result, 3Com would find an increased market for their hardware by creating this sort of environment.
The reality is that businesses will NOT spend money maintaining their old hardware, they will buy new gear. Hobbyiest (like my college living group with a network strung up with end of life products) and users will use and maintain this gear.
Why will this help sales? 3Com needs to realize that there is little market for old hardware, and this change would do nothing towards that. However, the techies often take home the old gear when it is replaced. As techies interested in toys, they would be more likely to recommend the purchase of hardware that they can play with at the end.
3Com benefits as the first mover, therefore they should do so.
Alex
but what about pulicity. Any company that does anything related to open source, it's touted as something that will help linux, and anything on the newswires about linux is instantly picked up by CNBC, C|net, slashdot, and eventually major local tv stations.
"Convincing the hardware makers will be the difficult part."
Maybe not.
I confess complete ignorance of the product in question, but I bet it wasn't built in a factory owned by a company called 3Com. Furthermore, the firmware (that's what we're really talking about) may very well have been outsourced as well, or just bought off the shelf. The best kept secret of the computer industry is that there are almost no American "hardware makers" at all. They're all just marketing companies.
Most of the stuff Cisco sells, for example, never physically touches a Cisco employee during its entire lifetime.
And Compaq stuff is sometimes triple or quadruple subcontracted. Their big factory in Austin is run by FIC, and in Germany they produce in the old East German Robotron factory. Tatung produces all European HP PCs in Holland, and the printer are make by Selectronics in Hungary.
Ect. Networking products are especially prone to be out sourced. Go to a big computer fair and visit the Taiwanese stands and just ask who really produces what. You'd be surprised.
Some companies just label complete products, some insist upon design changes, some do their own firmware development, but almost nobody with a recognizeable brand name goes out and actually makes their own PCBs (drilling holes in plastic -that's a Chinese job, as they say) let alone squabbling with the foundry sales guys about allocation.
There is a sort of apartheid in the computer business, mostly because the sophistication of the Asian side has increased so quickly. Maybe the open source community could help overcome it. I'm always suprised how people with the technical savvy of the average slashdot reader can be so ignorant of where the stuff really comes from.
The trick would be to find out who really designed the hardware in question and to contact him directly. In exchange for other freebies, they might be more than willing to talk.
I bragged about my Karma at a job interview but I didn't get the job.
#3 if the product doesn't work because you need new specs you have to go buy another one.
As long as all the companies don't release their old specs they are better off.
If you liked this thought maybe you would find my blog nice too:
The reason 3Com is not supporting the TotalSwitch anymore (and I have one too, it's a SWEET piece of hardware) is that they already had a competing line of products (Their SuperStack switches). Depending on who takes over development of the code patches: The issue is the 100bT, it's not duplex, though it was supposed to be duplex upon market emergence, and was to be patched when they (USR) got sucked into the 3Com conglomerate. Personally, I'd be willing to contribute money (some, I'm not rich), webspace (which I have in abundance), or whatever, to support a project designed to bring the product up to the promised specs, buy/develop the rights to the firmware code, etc.
Chas - The one, the only.
THANK GOD!!!
Chas - The one, the only.
THANK GOD!!!
It would cost very little for a company to release the documenation for obsolete hardware and allow the Open Source community to maintain it. The problem is they would lose sales for new products. Without the new sales, they can't improve the product line more, and they fall by the wayside. That is a bit extreme, but it is something that should be considered.
Companies discontinue products all the time. Usually it's because the product has lived it's useful life, and needs to be replaced, or their estimated profit margin wasn't fulfilled.
Not to be a conspiracy theorist, or anything, but I know sometimes companies discontinue a line of products so they can introduce a new line of similar products, and sell to their current customers who now have 'obsolete' products.
They've essentially just forced a bunch of clients to use unsupported gear, or pay someone(hopefully them) to replace it. If this is the case, open-sourcing would be counter-productive.
I don't see why this should be a difficult sell for any hardware manufacturer who's gotten out of stone-age source code jealousy. Why should they object to putting their boxes and their brand name on your desktop? The situation should be no different from the one in which Nike(TM) does everything it can to get its logo on the chests of millions, even if they're not athletes and say nothing about the quality of their shoes. If your IT people go to work and see "3COM(TM)" every day; if the administrators have the 3com(TM) name in front of them so that it's the name that pops into their head when its time to make more purchasing decisions; if they can generally get their name hard-coded into your product-inertia... how could they refuse?
Companies may not care about extending the life of products they no longer support. But extending the reach of their name -- that's something you can sell them on.
- Michael Cohn
-----
Go ahead, blame me... I voted for Nader!
A company like 3Com could agree to provide a certain number of ethernet switch ports in working order at your site for a monthly fee. They can update and upgrade the product at their leisure. If their equipment stops working for a previously agreed time period (maybe if it has less than 99.99% uptime per month), then you pay nothing that month. Once they replace your equipment, they take the old equipment and either recycle it or reuse it. This would be a more efficient system because every party has a strong incentive to avoid waste.
With this system, there would be no need to open the specs or source code of obsolete hardware, because companies would be offering free firmware and driver upgrades for much longer periods of time.
I am not a lawyer.
It would cost very little for a company to release the documenation for obsolete hardware and allow the Open Source community to maintain it.
I love recycling --especially the kind where hardware gets a new lease on life in ways we never dreamed.
Economic reasons? Think environmental. Think of the savings when our environment is concerned. Yeah, I know, most people don't give a damn about all that toxic lead used to solder those several layer circuit boards or worse yet all the chemicals needed to move them through hundreds of costly steps in production.
Let's promote green open source solutions and the companies that support them.
I would be asking 3COM why they are not supporting their own products, even if they were acquired in a merger, and if this indicates the level of support that customers can expect when current product lines are discontinued. Why should the customers be expected to take over the sustaining engineering? Even with source code and schematics, are the customers going to have the software and hardware tools needed to generate updates?
Mea navis aericumbens anguillis abundat
With things they own like the old network cards 3COM have been extremely good. I asked for documentation on the etherlink MC/32 and not only got documents back, but on paper. The times I've had problems with 3com docs have been when 3com dont own all the rights, when they are still filing patents on the hardware and it might cause them a problem that way.
Stuff like the TC boxes they probably don't own all the rights to. The code they use undoubtedly contains large licensed components and I can quite believe they dont _have_ good documentation except for the source to release.
Certainly when I worked for 3com rapops we had stuff inherited from Sonix that had basically -no- hardware documentation.
As vendors go 3com have been one of the most supportive to Linux, but I don't think you can expect them to do the due dilligence to release sections of code or go off and write docs for random dead junk switches. Maybe if you offer
to cover their costs for the process ? Do you love
the hardware enough to offer them $20K to do the work - remembering the HW may be too specialist to run any normal OS.
The /. community is mostly made up of IT support staff, a lot of SysAdmins and a handful of people that matter decision wise. All that needs to be done, assuming everbody can stop bickering and hold the same thought in their heads for more that 10 seconds, is to write a letter to 3Com, for example, and tell them that we're after the specs / source for their discontinued hardware, and if we don't get it, we'll all go off to Cisco, or other appropriate rival, next time we need something. It's called blackmail. Whenever said company releases the docs, we turn to Cisco and and tell them that now 3Com have opened up, you have to do the same or you can kiss our checkbooks goodbye. That bit's about market share. Now, all we need is somebody that can liason, is semi-responsible and doesn't mind pestering companies. I nominate CowboyNeal. Job done.
Argue the point all you want, but what needs to be done is getting the information for new hardware, as it's released. Only time and added pressure on the hardware manufacturers is going to make this a reality.
I/O Error G-17: Aborting Installation
The lifecycle of an at-market product is recognizable by four distinct phases:
General Availability: Selling is unrestricted in target markets. Carrier for new technology introductions. Marketing efforts to actively promote product. Resources allocated to enhancing and maintaining existing product.
Functional Stability: Product not targeted for new sales. Available to existing customers only, and resources allocated only to fixing major 'bugs'.
Maturity: This phase is often combined with Functional Stability. Sales are suspended, and existing fixes are made available, but no resources allocated to fixes. Limited support provided.
Retirement: Product is discontinued. Support, if available, is not dedicated, and often comes with surcharge.
Companies rely upon the product lifecycle to ensure that their Generally Available products are successfull. By extending the usefullness of products in the Retirement phase, GA products will be adversely impacted.
It is therefore not in the best interest of most companies to open spec hardware in the Retirement phase, nor is it benificial for a software company to open source Retired software products.
There may, however, be advantages to open spec or open source earlier in the product lifecycle.
-jerdenn
1. Said hardware contains design secrets used in current products.
2. Designs employ proprietary specs possibly used under license from other companies.
If I'm not mistaken, #2 is one reason why IBM is unlikely to ever opensource OS/2. Doesn't it contain some code written by MS?
It can't hurt to push for open hardware though. I would like to see these companies contribute something back to the community that has made them so wealthy.