Slashdot Mirror

That being said, I have read a couple of pieces coming from academia that really impressed me. They were primarily targeted at real problems. For example, I read a great piece on the now defunt LRP and some other stuff on clustering too. So there are some gems out there if you look for them.

I'm personally a fan of LinuxJournal even though it's not really an academic publication. They do some pretty interesting stuff and I usually get one or two good ideas per month. I do like reading what others are doing so that I don't have to repeat the process. The sad fact, however, is that many universities are behind the technology curve when it comes to leading edge research. Just because their school's students were the first to have MP3 players doesn't mean a university knows diddly about the current state of technology. I'd bet that there's more of interest on /. than what's happening in all the computer labs at all the large universities all across the country, trolls notwithstanding.

Of course they didn't make it quite as good as I had it laid out.

XML? I considered it and it's a bad idea. Too much work for machines OR humans to parse. They missed the boat on that. It also doesn't look quite as robust as my design either.Then again they're only a multi billion dollar company. What do you expect?

Maybe in another 15 years I'll actually get to see an OS something like what I was building in 2000. : P

Perhaps they were chosen to step on Matthew Grant's linux-based single-floppy distros "Eiger" and "Matterhorn".

Those were from the LRP, which is no more (LRP is dead, long live LEAF!)

Or maybe it's like Dilbert says, all the good code names have already been used (I myself am working on "project phlegm").

Look at the petty squabbles that break out in the development community. BSD vs Linux. XFree86.org vs X.org. Gnome vs KDE. We either put aside or quietly tolerate a lot of this nonsense to work together against Microsoft. A sort of "the enemy of my enemy is my friend" as it were.

I do not condone Microsoft's strong arm tactics, but neither can I say that we would be in the same or better place if Microsoft never was. I'd really like to believe we would, but experience tells me otherwise.

Look at LRP. It was an excellent idea. But there are so many "acceptable" alternatives there was no demand for it and no real support. Now, if M$ (or Cisco) had completely dominated the router market (no Linksys, NetGear, DLink, Belkin, etc) this project would have probably flourished simply because it was a better alternative. With no "arch villian" in the router market, it was just another alternative. Not a better one, just a different one.

Maybe Linux flourishes, in part, just because it's "not Microsoft". Would it get that attention if there wasn't a M$ to hate?

MS would be developing these bugfixes and developing the software whether they gave these million copies away or not.

Yes, just like the Linux router project.

IP has no value?

...the Linux Router Project, a floppy-based 386-compatible micro-distro which served as the basis for (among other things) Coyote Linux.

• Coyote Linux
  
• LEAF
  
• LRP (now dead)
  

We're talking about a dumb terminal here, aren't we? Let the user login with his card, enter a passcode, then enter input which gets sent to a server somewhere to be processed and which sends back either output to be displayed to the user or output to be read by the machine which gives you your money.

The same criticism applies to Diebold's voting machines.

This is why Linux would be such an ideal solution. No application of Linux has impressed me more than the (now sadly defunct) Linux Router Project, simply because it demonstrated how for many tasks most of the operating system amounted to nothing more than ballast. They were able to boot a router from a floppy.

This is how I think an ATM--or a voting machine--should work. The amount of software should be kept to an absolute minimum if for no other reason than that it minimizes complexity, and in these kinds of applications, complexity is the mother of all evil.

And in the case of the voting machines, it would also greatly assist in auditing the code and making sure that what you think is executing is what's executing.

In your init.d rc and rcS scripts add something like this to the start up execution case block:

[SK][0-9][0-9]+*) # Background
$debug "$@" & ;;

(Debian shown) and then rename start links S+[name].

The + will now denoted that that the process will be backgrounded at execution and will parallel your boot. You still need to make sure that dependancies are run first.

Of course whether you do this or use gmake this is still 20 year old mickey mouse bullshit, and the entire sysvinit system ought to be scraped!

def ShamelessBegging ():
Somebody give me money so I can write a modern core unix userland already!

Dave

It was only a few weeks ago, that the linux router project died through lack of funding - it was reported here.
Try not to find yourself in that hole.

LRP is the grand daddy of many "embedded" linux projects. LRP proved two concepts, 1) the need for GPL appliances that run from ram and essentially read-only media, and 2) a clever compressed read-only package system (.lrp instead of .rpm or .deb) for conserving boot media storage space. These ideas spawned LEAF, CoyoteLinux, and forshaddowed Knoppix, which all boot from floppy or CD-R media with compressed files to improve storage.

LRP was floppy firewall distro, that did not need a harddrive. It needed only 386 PC or better, 2 Nics, floppy drive, and sometimes a keyboard and monitor. It did not do fancy things, just NAT routing, firewalling and DHCP. But you could add .lrp packages for other cool features like DNS caching. The .lrp packages were just a renamed .tar.gz with binaries compiled a certain way, but they worked and saved space. Although building an LRP floppy was not easy for a novice, the package system made floppy firewall setup MUCH easier. With developers shrinking package sizes again and again, other lrp packages could be added, or log files could be added. Very clever.

But LRP failed to inivate fast enough, (e.g. I lobbied for a bootable CDs, to no avail) or document well enough, so Linux Embedded Application Firewall [LEAF] forked off. LEAF got space on SourceForge and spawned flavors, such as Oxygen, Dachstein, Eiger, Bering and others quickly helped fill out the space, improving core technologies and documentation. LEAF added bootable CDs and tons of packages. But LEAF struggled with picking a GlibC version and development of extensions became some what Balkanized.

The size limitation of the floppy made 2.4 kernal and iptables unatainable. Chuck Stienkhuler removed this boundry with his LRP-CD, which could fit every major linux ethernet driver, and so much more.

When I saw that, I thought, "well why not a full distro on a bootable CD", and was pleasently surprised by finding Knoppix. I even was the first person to mentioned it on Slashdot. [search Knoppix in stories on slashdot and find the first entry :) ]

LRP also spawned the CoyoteLinux firewall, which added a Win32 floppy build exe and a linux floppy build bash script. It makes building a floppy firewall really easy.

Death of LRP is not a surprise with LEAF on the scene. There is much life in the "embedded" linux space beyond firewalls. LRP got thing moving and many other GPL projects have adopted the core ideas and kept up the rate of acceleration. Bootable CD distros are exploding, into Mesh Networks, MAME systems, Linux on X-box hacks, PVR systems, LAN MP3 Servers, print server, LAN DNScache/DHCP/NTP server, Honey Pots and on and on. We will se more and more bootable CD distros, that will make our lives easier, and take the strain out of admin and system upgrade. Oh look, a new ISO on line, I down load and reboot my system. If it does not work, I pop the old CD-R back in. No muss, no fuss.

LRP is dead, long live LEAF and Knoppix, and ...

-Nathaniel
Mac Refugee, Paper MCSE, Linux wanna be.

Infact a 486 is overkill for a home based router. Its inet, xinit and other utilities that are highly bloated for a 486 which make it a little slugish. Try the linux router project. Its very tiny and designed for old systems to make them routers only. Infact it can run even without a hard drive! All you need is a floppy to load it and then forget it.

Speaking of old hardware I remember trying out Linux in 98 when I came to a Linux powered website powered by a 100mhz 486! No you did not misread that.

The webpage designer mentioned that his website gets thousands of hits a week and transfers gigs of traffic! Besides using the 486 to serve webpages the author benchmarked it with apache and the linux 2.0 kernel with a pentiumpro running IIS.

The linux 486 machine creamed the IIS pentiumpro by a significant margin. I believe both systems only had 48 megs of ram so that might of been the bottleneck on NT. Ram was alot more expensive and NT + IIS was a beast. NT back then typically ate more then half of a systems ram. I think he also was playing with perl apache modules to make the linux box faster. I do not remember for sure. Perl could not be loaded as an isapi app back then and they could only use cgi on IIS. My last statement is just speculation of course but I can otherwise explain why apache creamed it.

Keep in mind many older cisco routers use old motorola 68030 processors which barely have 386 level speeds. The internet backbone used these for years and they can transfer enormous amounts of data. They do not need to run quake3 or compile Mozilla. They are fine at what they do.

Well, one good reason is if you are using your last 386 as a home router / server. The ancient Linux Router Project is still up on the web, for those who want to use a 386 in this manner. There was even a Slashdot article on this last year, though for the life of me I cannot find the article in order to link to it. I'll keep trying, but wouldn't mind a gentle reminder if someone else knows...

Blacklisted

The problem is, if you look at the linuxrouter.org main page, you'll find that the site hasn't been updated since May 3, 2001. Most LRP development these days is on the LEAF site.

You cant afford $60? Or your want a real router?

Google/Linux router floppy gives Linux router project

Have you ever tried LRP (Linux Router Project)? I suppose no, because you don't get the "NFS mess" you talk about. I've seen it running from floppy disk and compact flash card, and it worked great.

Sure, you don't boot in 3 seconds when you do it from the floppy disk, put the flash card was a better anyway...! and you don't need to flash your BIOS everytime you want to upgrade...

There are a lot more of minimalist Linux distro that don't have this "NFS mess", like tomsrtbt and Small Linux... go to Linux.org to see a huge list of minimalist Linux distro! (of course, you need to select "minimalist" in the category box...)

...which you will inevitably do...

I do not think this word means what you think it means. :)

Considering this runs off a single floppy disk, you may have trouble squeezing all those other daemons on it, even if you use bzip.

Linux Router Project

since there are over a 1000 posts, sorry if this is repeated.. i use a floppy for a mini distribution called toms [toms.net]. its fun to play around with. also linuxrouter [linuxrouter.org] is another distro that fits on a floppy and as the name implies is great as a router.

256M CF cards aren't as cheap, but you can fit a pretty decent OS on one, or most of a compressed boot partition.

A decent OS will fit on far less.

My router/firewall/NAT box/IPSEC VPN/DNS server boots FreeBSD off a 64MB card on an IDE converter. Boots really fast now ;)

And doesn't LRP fit on a floppy?

The only problem with using CF for removable media is the limited number of write cycles, but I suppose the same could be said about floppies. I wish LS-120 has caught on; that was a kick ass drive.

LRP's project might be of some use. Yes, it means getting a cheap box to run Linux, but then you can use all of that real neat networking software that's available for Linux boxes but isn't available for Windoze boxes.

Plan ahead. Change MS Office default file formats to friendly open formats now. Setup all servers to linux. Test users interest with Knoppix bootCD-OS. If Knoppix goes well with users install Win32 versions of favorite Open Source apps. Run your own audit of the company's win32 systems. Duplicate and convert propietary data files to open files (e.g., .doc to .rtf) by hand if you have to, but some folks may have automated it. Find special case win32 software needs, and see if wine will support. Then convert several users at a time, starting with the tech savy, and see if you can avoid mutiny. People have invested a lot of time learning one way, they hate upgrades, (remember the last of many Microsoft and Adobe upgrades?) so be patient when you upgrade your users to Linux desktop.

Changing MS Word users default Save file format to .rtf is easy. It will make all upgrades to OpenOffice much easier, and allow several version of MS Office to play nice, even if you don't upgrade. Years later you will be able to read old MS Office files, hooray!

As many have said:
Setup servers for windows file and print, web hosting, DNS, DHCP, and SMTP (samba, apache, bind, DHCPd and sendmail) in the back office. LEAF, LRP and CoyoteLinux firewalls are an easy place to start the conversion.

Try Knoppix BootCD-OS (debian) on every box, see if users can deal. It is complete Desktop with OpenOffice, Xmms, ogg-vorbis, Gimp, FreeCiv, and tons more on 700Mb CD-R. It auto-detects a ton of hardware, such as sound at each boot, and does not get installed to harddisk. It needs 128Mb Ram, or pagefile/swapfile/scratch disk on a box with less ram. If the user can't deal, eject the CD, and reboot back to MS Windows.

Setup each Win32 computers to run a script stored on a central server, at each boot. It saves a ton of work later.

Getting Win32 users into the OpenSource thing by installing Win32 OpenOffice and Mozilla on your current MS Windows install base. See how that goes with the users.

Convert your existing data from .rtf to .doc is critical. The user is a tease, no email or code! But it is an idea that should be packaged.

Run an audit on your Win32 systems. Get a file dump e.g., "dir /AH /ON /S > m:\filetreedump\box2tree.txt" on Win98, and goto to regedit and dump the registry to text file e.g., m:\filetreedump\box2reg.txt. Someone needs to write a nice perl based evaluation tool to audit what apps and software keys everyone in the Windows network is running.

Converting data is essential. Collecting data from users computer and registry, and inserting it into new email client, and Office apps should be automated, but no one has done it yet.

Wine testing for special apps is important.

After careful planning start rolling out conversions. This way you can convert data, support all the apps, and not loose users.

-Nathaniel

For a firewall sitting between my LAN and my cable modem, I use LRP. Runs on anything from a 486 upwards, off a floppy. Once you've finished configuring the floppy (takes between 5 mins and 3 hours, depending on how experienced you are with Linux and networking), you can simply write-protect it, and you have a completely uncrackable system. If someone breaks into it, just reboot the computer and it'll load everything back into ramdisk. It can also act as a dhcp and dns server. Check hereand here for documentation.

I wrote what was once widely appreciated as the most useful howto for using LRP. It is now woefully out of date, and I recommend Eigerstein or Dachstein, which are so well-designed that they don't need that kind of detailed documentation.

I can shed a little more light on the middle-recent history of LRP and LEAF. Two years ago, LRP was indeed the center of all linux floppy firewall/router activity. However, people were starting to innovate, and Dave Cinege (who owns the domain name) never seemed to find the time to update his own work or incorporate that of others. It was a running joke on the mailing list. It would not have been much work for Dave to at least put up links to the sites documenting and extending LRP, but it never seemed to happen.

For a while, linuxrouter.sourceforge.net (now changed to leaf.sourceforge.net) was a repository of all the extra work. Before that everything had been on a crazy collection of obscure personal websites (like mine).

Dave promised major updates to LRP, and then gave up on LRP and decided a completely new, cool project was necessary. This was around the time Tim McVeigh was executed, which Dave considered the murder of a hero or prisoner of war. Without getting into politics or morality, I merely note that it was the last straw for many people, who made a complete split and formed LEAF. I presume it was the rancor behind this split that keeps Dave from mentioning LEAF on his website.

Unfortunately, if you type "linux router" into Google, LEAF shows up way down the list -- maybe 20th.

IMHO, the people working on LEAF are dedicated and impressive. It remains far and away the best floppy-based router/firewall available. It is certainly the most actively maintained.

I wrote what was once widely appreciated as the most useful howto for using LRP. It is now woefully out of date, and I recommend Eigerstein or Dachstein, which are so well-designed that they don't need that kind of detailed documentation.

I can shed a little more light on the middle-recent history of LRP and LEAF. Two years ago, LRP was indeed the center of all linux floppy firewall/router activity. However, people were starting to innovate, and Dave Cinege (who owns the domain name) never seemed to find the time to update his own work or incorporate that of others. It was a running joke on the mailing list. It would not have been much work for Dave to at least put up links to the sites documenting and extending LRP, but it never seemed to happen.

For a while, linuxrouter.sourceforge.net (now changed to leaf.sourceforge.net) was a repository of all the extra work. Before that everything had been on a crazy collection of obscure personal websites (like mine).

Dave promised major updates to LRP, and then gave up on LRP and decided a completely new, cool project was necessary. This was around the time Tim McVeigh was executed, which Dave considered the murder of a hero or prisoner of war. Without getting into politics or morality, I merely note that it was the last straw for many people, who made a complete split and formed LEAF. I presume it was the rancor behind this split that keeps Dave from mentioning LEAF on his website.

Unfortunately, if you type "linux router" into Google, LEAF shows up way down the list -- maybe 20th.

IMHO, the people working on LEAF are dedicated and impressive. It remains far and away the best floppy-based router/firewall available. It is certainly the most actively maintained.

Was there any reason for this possibly very damaging statement?

Yeah, because at the linked site:

• There have been no releases since 0.9.8 on 12 Sep 2000 (a year and a half).
• The only news since then has been three seperate sponsers (Cyclades, VA, and Sangoma). It's not clear what the money is being used for.
• The mailing list archives, give 404s on the -devel list. Only the users list seems to be active.
• The "unstable" directory on the site contains only (besides the 0.9.8 release) a few kernel patches made to 2.2.19 in July of 2001.

On the other hand, this site seems quite active. I'm not sure what their relationship is.

Was there any reason for this possibly very damaging statement?

Yeah, because at the linked site:

• There have been no releases since 0.9.8 on 12 Sep 2000 (a year and a half).
• The only news since then has been three seperate sponsers (Cyclades, VA, and Sangoma). It's not clear what the money is being used for.
• The mailing list archives, give 404s on the -devel list. Only the users list seems to be active.
• The "unstable" directory on the site contains only (besides the 0.9.8 release) a few kernel patches made to 2.2.19 in July of 2001.

On the other hand, this site seems quite active. I'm not sure what their relationship is.

You can use an old PC with no hd and two 10/100NICs. Linux Router Project should fit the bill.

--Mike

I use Freesco on a DX2/66 with 16mb (no HD). Remote interface through thttpd, DHCP server, caching DNS, support for DynDns, time server, multiple NIC's and/or diald and dialup, all on a 1.44 floppy. By default, the logging is not very robust but it works fine for what I am doing.
You can do similar with the LRP

Not that the method the story describes is not good, just there are configurable alternatives to balance security and ease of use.

Why windows does not run off a ramdrive. I mean, modern PCs all have at least 512MB ram, why not load up Windows once, and then never access the disk drive again?

In fact, many if not most minimalistic Linux distro's do this. Specifically, Linux Router Project do this. I use it extensively. The kernel boots and the file system decompresses to a RAM drive. It's very fast.

CompactFlash storage has an IDE interface (more or less). There are cheap adapters to use CF as an IDE device on any motherboard - intended for embedded systems. The adapter has no components on it - it's purely a wiring harness for the different sized plugs, and a jumper for master/slave.

This shows as a small harddisk in your BIOS, and anything that believes the BIOS should work, and boot from it. I just got one, and started playing with it for a couple of applications (silent X Terminal, firewall), and so far I have DOS booting really quickly, and LRP almost working.

I could see a pretty decent little business in doing diskless firewalls and routers using an LRP image or an EmBSD firewall image on a microATX x86 box. Just add the configuration files to your keychain, boot it and run in RAM. Sweet.

The problem is hubs. I have yet to see a good gigabit hub for under $2k or so. Most of the gigabit-compatible hubs offered use gigabit for uplink, and a handful of 100-base-T links for the rest of the ports.

What sort of crack are you on? Check out the Linux Router Project , freesco, etc. I own an ISP that is Cisco free using entirely LRP based routers and firewalls.

Some of these firewalls and routers are 486 and Pentium based systems. The core router currently is a measely Pentium 75 with 16 MBs ram and all is does it route packets....with ease. The core router is connected via fast ethernet to the backbone of my provider where I'm colo'd. The inside fast ethernet connections go to various class C networks.

Aside from the more typical ISP functions, we light office buildings and sell bandwidth on a private network to networks and individuals within the building, all connections firewalled. There Pentium based machines with 16/32MB RAM and 6 ethernet/fast ethernet ports and K-6/2 or Celeron based machines with 64/128MB of RAM and up to 17 fast ethernet ports or 8 T-1's w/ CSUs. Most of these are rackmounts and some are "obsolete" low profile business grade systems.

Using a PC as a router in place of a hub isn't an option, as one gigabit ethernet card will come very close to saturating a 32-bit 33 MHz PCI bus. Start streaming large amounts of data through the house and the router will fail to handle the traffic.

I think for a router you will find that in practice it will be pretty difficult to saturate the 32-bit 33MHz PCI bus. This is a home network and we have a hard time saturating the bus in 100,000 square foot (12,200 square meter) office buildings with several hundred people getting internet access. Also, if you really must use GB Ethernet, I've seen copper Gigabit Ethernet cards for under $50. At that price, skip the hub/switch and put multiple in the router and bridge them, for connection to the downstream connections. If performance is critical, skip the hub idea altogether and really route or get a switch.

For the time being, Gigabit Ethernet is probably unnecessary so use fast Ethernet. You can even get a 4 port PCI card that is a fast Ethernet switch. So that and another fast Ethernet card, and you've got your router. Or put 2 fast Ethernet cards in the router and buy a switch. Even the Realtek 8139 based cards work just fine and can be under $10 with some looking.

At home, or at work, a PC based router is an elegant, inexpensive, highly reliable router.

I'd imagine that either OSPF, or possibly BGP, would work just fine for this sort of thing. The problem is, you need a REAL router to do these things, not one of those $200 USD Router-In-A-Box jobbies that you can pick up from D-Link or Linksys. And even then, you're not going to be able to do that over DSL or Cable, because I don't know of ANY ISP that will provide a BGP table over those connection types.

More likely would be some sort of setup where their would be a central node that those gateways would report to, and the routers at those gateways would report throughput for their links back to the central node. That node would then distribute that bandwidth as equally as possible on a per-request basis.

This basically would be just like setting up any kind of NAT-based network with more than two connects to the internet, but without the benefit of BGP to help things along. It's possible, but sticky.

There's also the subject of the real routers. Netopias won't cut it here either; you'd almost have to have a Cisco. Alternately, you could set up Linux boxen to serve as your router; Zebra is supposedly pretty far along, and would work for internal traffic distribution. There's also LEAF and LRP, two closely related projects for doing a single-floppy router/firewall/NAT device. Find them at:

The LEAF Site

Or:

The LRP Site

It'd take a lot of effort, but if you happen to live in a high-geek-ratio neighborhood and you can share the implementation efforts across other shoulders, it should be easy enough.

>Is there any "real" use

The fact that there are actual shipping products that use Linux in embedded devices indicates that there are "real" uses. Try reading the article.

>Yes, but linux is a rather "large" OS.

I would tend to disagree with this.

Distributions of Linux tend to be very large. The size of the "operating system" is a very variable thing.

Of course, the kernel itself can be built to suit and can be made quite small. If you are building an embedded device, the list of stuff (outside of the kernel) you don't have to package on your system would trim down the size considerably:

no source code
no development libraries
no development tools
no X

probably only the one application for which the device exists instead of the hundreds that included in a typical distro.

"Linux" can be made small enough to fit on a floppy disk and run completely on a ramdisk (LRP).

My example is my 386 with 4 meg and 100 MB drive running my cable modem' masq box. I'm running a kernel with everything stripped out but the bare essentials. I'm using Debian with just the barest set of packages installed. This machine is a single-purpose device with a very small OS. If I had the time or needed to, I could probably make this even smaller.

The beauty of using Linux for these purposes is that you can trim it down to just the functionality to want/need to get it to fit into your device. At least, doing so is a lot cheaper than rolling your own OS.

Okay, sorry, I advocated the quiet, cheap solution that can be made out of spare parts, for a low cost, doesn't have any moving parts save for the floppy drive, and is endlessly upgradeable and configurable(LRP).

I must apologize, I guess the commercial solution was better. When I posted, I didn't think that this guy needed some more information about a PC-router solution, that maybe he didn't know they could be run fanless and without a hard disk. I didn't think that he might not have known that there was specific distros for this purpose. What I did think was "This guy wants hardware, lets piss him off." This guy mentioned $$$, this solution was cheaper then he wanted, but that is obviously a drawback. He also mentioned wattage, which the solution I gave used little. But I suppose your reading skills aren't the best either, are they?

I wouldn't condemn others so easily, if I was you. There was redundant posts, I agree. But a PC-hardware based solution is just as competitive as the low-end custom router solution, if not more so.

get an old PC and setup a router with help from the Linux Router Project then you can get fancy and have all the features you'd ever want, throw a disk in it and it doubles as a file server, cache dns lookups and web content.

somehow it's more fun that way....or is it just me that thinks this Linux stuff is fun?

Have a look at the linux router project (lrp). http://www.linuxrouter.org. I have had it running 24/7 for about 6 months now, and not once has it crashed (not surprising, since it's based on linux). However, it also runs directly off a floppy, which means the PC you run it on is virtually silent.

I have it running on a 486-66, 16MB, no hdd, to connect my cable modem to my LAN. Of course, you can also use it with Tx/DSL/ISDN/analogue.

Sorry, this reads like an ad, but I really love this distro - it has made life so much easier.

I don't know if someone said that already, but there's a very cool german linux-router project (xDSL/ISDN) focused around: fli4l
From fli4l.de:
"Fli4l is a single floppy Linux-based ISDN, DSL and Ethernet-Router. You can build it from an old 486 based pc with 16 megabyte memory, which is more than adequate for this purpose. The necessary boot-disk can be built under Unix, Linux or Windows. You don't need any specific Linux-knowledge, but this would be useful. You should have some basic knowledge about networking, TCP/IP, DNS and routing though. For extensions and further development, that exceed the standard configuration, you need a working Linux-system and Unix/Linux knowledge."
Personally I use it as a DSL Router for my 2 PCs and it works great: F***in fast, secure and easy. If you want a ISDN/DSL Router check it out., otherwise LinuxRouter ist the better choice.
X

I'm currently using an old 486SX25 that my employer had slated for the dumpster, with 2 ne2k nics that are from old PC's also. It has 12MB of memory and NO HD. Checkout www.linuxrouter.org. Help save the landfills and use them ol PC's.

The project is excellently documented, support is widely available on websites and newsgroups and setup and security is not much more difficult than in an average Linux distro (command prompt though).

The only problem I encountered was the recognition (and subsequently configuration) of network interface cards on my old 486 box. But with plug 'n pray capable Pentium systems that should not be an issue anymore since you can have the NICs recognised in 'normal' circumstances first, copy the settings and use them in your router setup.

Succes!!!

You don't need a hard drive for a firewall/router made from an old machine. Check out the LRP for a solution that fits on a single 1.44 mbyte floppy that can be write-protected and just needs to be power-cycled to be reboot.

The linux router project is one of the best sources of info on getting that old 486 to work as a router. I had mine running fine until about two months ago when I was able to get a Netgear router for $30 (easier for parents as I was leaving for college).

See www.linuxrouter.org for more information.
Steinkuehler's EigerStein was the distro I used - worked very well.

-Doughnuthole

Try the Linux Router Project. Designed to fit on a floppy and then expand to a RAMdisk. It should work with just about any PC you've got sitting around.

This thing is huge. I was thinking of something around the size of a 5 port hub for $50. If it's as big as a computer why not do it in software a la Linux style? Hell, you could probably get away with writing a couple of shell scripts to generate traffic and some deliberately bungled iptables mods. Put that on a Linux Router Project floppy disk and your done. Oh, wait a minute. I wonder selling a big box with little chips inside is more profitable? Hey!

This involves mounting a floppy with tar gzip files, then copying those into ramdisk. This happens late in the boot process (see the script "linuxrc" (ie, pkgsrc/root/linuxrc). Should work with a removable drive.

Besides LRP and Freesco, how about digital cameras? Let's see - Panasonic manufactures the only digital camera that can store hundreds of thousands of images on a single floppy. Sounds pretty cool to me! Of course, you have to have a compatible drive in your computer to read the disk, so Panasonic makes sales in both areas.

Sure, there's a Linux driver and all that jazz, and it works fairly well, etc. etc., but M Systems has a reputation in the Embedded Linux community for wanting you to sign your soul over before being able to use their stuff. The license is HIGHLY proprietary on their Disk-On-Chip product, and I'm sure that something similar will hold true for their other products.

If you aren't a zealot when it comes to licensing, then you've got a pretty decent USB storage device here. If you're a stickler for non-proprietary systems, then avoid it like the plague. The Linux Router Project and the LEAF Project have been looking for a way to incorporate something larger than the floppy for embedded firewall/router boxes, and M Systems' stuff was looked at and discarded due to the licensing issue.

Stuff to be aware of. Why not get yourself a cheap IDE-based Compact-Flash card reader if you need the space? They've got them for around $20 plus flash, and with flash being used in digital cameras, you can get some pretty decent-sized ones for cheap.

If I can replace my routing box with something that's completely fanless(even w/o a harddrive)

Like the other reply says, you could just get a Linksys router. The other option would be to take your router box and use a LRP disk to boot from. LRP boots from that disk (1 or 2 3.5" floppies), works well on a 486 with at least 12MB RAM. With a slower 486 (33MHz or less) or a large heatsink, you could likely get away without a CPU fan and you certainly wouldn't need the hard drive. As long as you don't use too much power, you could probably even remove the fan from your PSU.

I've got a p100 in my room in which I recently installed a 45GB IBM DeskStar to replace the older drives(2.1G, 1.2G, 850MB). The CPU fan actually makes the worst noise pollution since it's so low (it resonates through the case and floor, which just work to amplify it). I'm also considering installing a LRP disk in it to be rid of the 250MB boot drive which is rather flakey.

Anyway, just another suggestion.

It sure would be nice to substitute a shell script or a small C program for that Perl script. Many folks run full Linux distros on their desktop and relegate the firewalling duties to small routers running something like LRP without software packages as huge as Perl.