So does linux. Linux can be: A packet forwarder (router) A packet filter (firewall) An IP masquerading packet filter (NAT firewall) Can run RIP, BGP, EGP, OSPF (via gated) Maybe I'm being naiive here, but what does Cisco offer beyond this (besides the availability of higher performance)? --Adam -----Original Message----- From: Michael Shields <shields@crosslink.net> To: dirk@power.net <dirk@power.net> Cc: nanog@merit.edu <nanog@merit.edu>; list@inet-access.net <list@inet-access.net> Date: Wednesday, October 28, 1998 4:01 PM Subject: Re: Linux Router KIT In article <19981028100541.00359@orlando.power.net>, dirk@power.net wrote:
Linux doesn't just kill Microsoft's NT and Solaris. It also eats Cisco for lunch.
This isn't true. IOS does a lot more than just get packets from interface A to interface B. (In terms of managability as well as functionality) -- Shields, CrossLink.
Lots of types interfaces? EIGRP? Things that Linux/PC does have is Hard drives that fail, etc; I guess you could do a flash thing, though. Other protocols, like IPX, Dec, ATalk, etc are not going to happen. Also, for a long time, Linux had a hard time with lots or routes. If you are going to use a PC as a router, use an Unix that has real net code, like xBSD. On Wed, 28 Oct 1998, Adam D. McKenna wrote:
So does linux.
Linux can be:
A packet forwarder (router) A packet filter (firewall) An IP masquerading packet filter (NAT firewall) Can run RIP, BGP, EGP, OSPF (via gated)
Maybe I'm being naiive here, but what does Cisco offer beyond this (besides the availability of higher performance)?
Linux doesn't just kill Microsoft's NT and Solaris. It also eats Cisco for lunch.
This isn't true. IOS does a lot more than just get packets from interface A to interface B. (In terms of managability as well as functionality)
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- ISPF, The Forum for ISPs by ISPs. October 26-28, 1998, Atlanta, GA. Three days of clues, news, and views from the industry's best and brightest. http://www.ispf.com/ for information and registration. Atheism is a non-prophet organization. I route, therefore I am. Alex Rubenstein, alex@nac.net, KC2BUO, ISP/C Charter Member Father of the Network and Head Bottle-Washer Net Access Corporation, 9 Mt. Pleasant Tpk., Denville, NJ 07834 Don't choose a spineless ISP; we have more backbone! http://www.nac.net -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
On Wed, 28 Oct 1998 alex@nac.net wrote:
Lots of types interfaces? EIGRP? Things that Linux/PC does have is Hard drives that fail, etc; I guess you could do a flash thing, though. Other protocols, like IPX, Dec, ATalk, etc are not going to happen.
Eh? IPX : http://sunsite.unc.edu/mdw/HOWTO/IPX-HOWTO.html Decnet : http://linux.dreamtime.org/decnet/ Atalk : http://thehamptons.com/anders/netatalk/
Also, for a long time, Linux had a hard time with lots or routes.
No longer applies. In fact Linux is now faster than BSD up to about 60-70,000 routes. BSD is faster at about 200,000. In between its about even.
If you are going to use a PC as a router, use an Unix that has real net code, like xBSD.
Please, leave religion out of the discussion... -Dan
On Wed, 28 Oct 1998, Dan Hollis wrote:
Also, for a long time, Linux had a hard time with lots or routes.
No longer applies. In fact Linux is now faster than BSD up to about 60-70,000 routes. BSD is faster at about 200,000. In between its about even.
Where/how are you doing simulations with that many routes? Feeding full routes to a linux 2.0.3x box running gated, I get: # cat /proc/net/route | wc -l 54677 and that command takes 13s to complete. When dealing with that many routes on a *nix box, better tools would be nice. On my 3640, I can do a show ip route blah, and get immediate response. On the linux box above, I have no such tool. I should fire up a copy of gated 4.x and see how gii performs, but its kind of a drag not being allowed to use the newer code on anything but personal testing/development systems. ------------------------------------------------------------------ Jon Lewis <jlewis@fdt.net> | Spammers will be winnuked or Network Administrator | nestea'd...whatever it takes Florida Digital Turnpike | to get the job done. ______http://inorganic5.fdt.net/~jlewis/pgp for PGP public key____
I should fire up a copy of gated 4.x and see how gii performs, but its kind of a drag not being allowed to use the newer code on anything but personal testing/development systems.
I contributed some of the "show bgp" GII code to gated, and when I used it on Digital's border routers (Alphas running Digital UNIX) it was invaluable. "netstat -nr" is nice, but looking into the routing process and seeing what choices were available was invaluable in answering some BGP peering questions. Stephen - ----- Stephen Stuart <stuart@tech.org>
On Thu, 29 Oct 1998, Jon Lewis wrote:
On Wed, 28 Oct 1998, Dan Hollis wrote:
Also, for a long time, Linux had a hard time with lots or routes. No longer applies. In fact Linux is now faster than BSD up to about 60-70,000 routes. BSD is faster at about 200,000. In between its about even. Where/how are you doing simulations with that many routes?
These are just numbers reported by Alan Cox and Alexey Kuznetsov.
Feeding full routes to a linux 2.0.3x box running gated, I get: # cat /proc/net/route | wc -l 54677 and that command takes 13s to complete.
Well thats great for benchmarking procfs B) I wonder how long 'netstat -rn' takes on *BSD with 54k routes. -Dan
On Fri, 30 Oct 1998, Stephen Stuart wrote:
I wonder how long 'netstat -rn' takes on *BSD with 54k routes. 6 seconds (output to /dev/null) on a Pentium something-or-other running BSD/OS 3.1.
How about 'netstat -rn | wc -l'
% time eval 'netstat -rn | wc -l' 53140 6.95s real 3.88s user 2.98s system Stephen
On Thu, 29 Oct 1998, Jon Lewis wrote:
Where/how are you doing simulations with that many routes? Feeding full routes to a linux 2.0.3x box running gated, I get:
# cat /proc/net/route | wc -l 54677 and that command takes 13s to complete. When dealing with that many routes on a *nix box, better tools would be nice. On my 3640, I can do a show ip route blah, and get immediate response. On the linux box above, I have no such tool. I should fire up a copy of gated 4.x and see how gii performs, but its kind of a drag not being allowed to use the newer code on anything but personal testing/development systems.
I agree, this is a big gaping hole in functionality. This has to go into gated because only gated has useful information like what the AS path for the route is, etc. -- Jawaid Bazyar | Affordable WWW & Internet Solutions Interlink Advertising Svcs | for Small Business bazyar@hypermall.com | 910 16th Street, #1220 (303) 228-0070 --The Future is Now!-- | Denver, CO 80202 (303) 228-0077 fax
Lots of types interfaces? EIGRP? Things that Linux/PC does have is Hard drives that fail, etc; I guess you could do a flash thing, though. Other protocols, like IPX, Dec, ATalk, etc are not going to happen.
Will you please point out other router than cisco which has EIGRP ? What kind of interface do you want ? You have async (multiport async), sync, ethernet, fddi and now atm is coming. BTW, do you get arcnet with cisco ? :-) IPX and appletalk have been there for a _long_ time. There is also a Linux DECNET project.
Also, for a long time, Linux had a hard time with lots or routes.
Define lots. You want full BGP table in a PC router ? Why :) ? Without any claims that it's a lot, our machine has 450+ routes, does BGP and OSPF with 3 NICs, constantly handles over 1Mbps data and it is a 486/66 with 16M RAM. A 2500 died when it was flooded with small packets, linux survived. I'm sure the box would still be ok when the load will double (probably won't see the day, though :)
If you are going to use a PC as a router, use an Unix that has real net code, like xBSD.
*shit. You have no clues. Linux was better at networking than BSD even in 1.2.x days ... All the limitations of the Linux/PC router are due to the PC hardware architecture. As seen on the list, people put 8 cards in the same PC. This exceeds the bus speed of a PC. Even a single 100Mbps NIC kills the PCI bus in most PCs should it run full speed. Also, you have to be very carefull with the NIC you choose. PCs simply were not built for forwarding packets and fast I/O. Of course a Linux/PC will never beat a cisco :-) but the cost is sometimes an order of magnitude lower for roughly the same performance. -- Matei CONOVICI, cmatei@roedu.net
On Thu, 29 Oct 1998, Matei Conovici wrote:
Lots of types interfaces? EIGRP? Things that Linux/PC does have is Hard drives that fail, etc; I guess you could do a flash thing, though. Other protocols, like IPX, Dec, ATalk, etc are not going to happen.
Will you please point out other router than cisco which has EIGRP ?
My point exactly.
What kind of interface do you want ? You have async (multiport async), sync, ethernet, fddi and now atm is coming. BTW, do you get arcnet with cisco ? :-)
HSSI?
IPX and appletalk have been there for a _long_ time. There is also a Linux DECNET project.
Great, another *project*.
Also, for a long time, Linux had a hard time with lots or routes.
Define lots. You want full BGP table in a PC router ? Why :) ?
Isn't that the crux of the conversation here?
*shit. You have no clues. Linux was better at networking than BSD even in 1.2.x days ...
Heh. Ok. (as in, *yeah right*).
All the limitations of the Linux/PC router are due to the PC hardware architecture. As seen on the list, people put 8 cards in the same PC. This exceeds the bus speed of a PC. Even a single 100Mbps NIC kills the PCI bus in most PCs should it run full speed. Also, you have to be very carefull with the NIC you choose.
Thus, omre reason to not use a PC for routing.. This was hashed and rehashed not to long ago on some mailing list (inet-access?)... Whay again?
PCs simply were not built for forwarding packets and fast I/O.
Again, thanks for agreeing.
Of course a Linux/PC will never beat a cisco :-) but the cost is sometimes an order of magnitude lower for roughly the same performance.
Not since the 2600 and 3600 have been released. -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- ISPF, The Forum for ISPs by ISPs. October 26-28, 1998, Atlanta, GA. Three days of clues, news, and views from the industry's best and brightest. http://www.ispf.com/ for information and registration. Atheism is a non-prophet organization. I route, therefore I am. Alex Rubenstein, alex@nac.net, KC2BUO, ISP/C Charter Member Father of the Network and Head Bottle-Washer Net Access Corporation, 9 Mt. Pleasant Tpk., Denville, NJ 07834 Don't choose a spineless ISP; we have more backbone! http://www.nac.net -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
Will you please point out other router than cisco which has EIGRP ? My point exactly.
Ah, so you are using _only_ cisco in your network ? Tough!
What kind of interface do you want ? You have async (multiport async), sync, ethernet, fddi and now atm is coming. BTW, do you get arcnet with cisco ? :-)
HSSI?
I'm sure that the moment someone will have a reasonably priced card for a PC and make specs available, linux will support it (probably BSD as well).
IPX and appletalk have been there for a _long_ time. There is also a Linux DECNET project. Great, another *project*.
... for a protocol that everyone *uses* :-)
Also, for a long time, Linux had a hard time with lots or routes.
Define lots. You want full BGP table in a PC router ? Why :) ?
Isn't that the crux of the conversation here?
No ?! Point being that a PC router (eventually running linux) can be more than successfully used as a low-to-middle end router, at very good value for money. Anything more than this looks a bit absurd to me ... If I have the money to pay for multiple transit providers, I most certainly have the money for a 4500 or more (as in real router) ... And BTW, I didn't imply that a linux box has troubles with many routes, but I never tested it myself. But once I read the source code for route.c and I don't think the number of installed routes is an issue.
... Thus, omre reason to not use a PC for routing..
PCs simply were not built for forwarding packets and fast I/O. Again, thanks for agreeing.
But I don't :-) I just think they should not be used for high-performance stuff, but perform just fine as long as you know what to expect.
Of course a Linux/PC will never beat a cisco :-) but the cost is sometimes an order of magnitude lower for roughly the same performance. Not since the 2600 and 3600 have been released.
Fine, replace "an order of magnitude" with "a lot cheaper" and you're still close enough :-) However, we're beating a dead horse. I think we both realize what can and can't be done with a PC router. It's just that you were overly criticizing Linux as a router without being at least informed and I felt an urge to react because I'm a happy linux user :-)) Cheers, -- Matei CONOVICI, cmatei@roedu.net
On Thu, 29 Oct 1998 alex@nac.net wrote:
What kind of interface do you want ? You have async (multiport async), sync, ethernet, fddi and now atm is coming. BTW, do you get arcnet with cisco ? :-)
HSSI?
Yep, available for pc routers running *bsd or linux.
alex@nac.net wrote:
All the limitations of the Linux/PC router are due to the PC hardware architecture. As seen on the list, people put 8 cards in the same PC. This exceeds the bus speed of a PC. Even a single 100Mbps NIC kills the PCI bus in most PCs should it run full speed. Also, you have to be very carefull with the NIC you choose.
Thus, omre reason to not use a PC for routing..
This was hashed and rehashed not to long ago on some mailing list (inet-access?)... Whay again?
They hashed it out and got it wrong? wow. Choosing the wrong NIC card and failing to make a PC do routing is not a valid reason for those who choose the right NIC cards to avoid a PC. Multiple 100 Mbps ethernet ports on many routers exceeds the capacity of the router. While many do have a total capacity equal to the sum of their ports, many do not, and it is not a general requirement to be classified as a router. Anyone contemplating any router should determine their capacity needs and judge it accordingly. Just because you might need, or just happen to have, 100 meg NIC cards does not mean you intend to use the total speed on all cards, or even a single one. A PC may well be inappropriate for maximum capacity routers, but they are usually chosen when the needs differ. In my case the total capacity I do need is less than 10 meg total. I won't be overloading a PCI bus @ 33 MHz.
PCs simply were not built for forwarding packets and fast I/O.
Again, thanks for agreeing.
PCs simply were not built for anything in particular, besides maybe Windows. Using them for anything else is adapting. But the world is full of people who do well at adapting things. If you know how to make it work, great. If you know you need to get more info to make it work, OK. If you can't make it work for what your needs are, then obviously it's the wrong choice. OTOH, high end routers are totally overpriced for low end uses, too.
Of course a Linux/PC will never beat a cisco :-) but the cost is sometimes an order of magnitude lower for roughly the same performance.
Not since the 2600 and 3600 have been released.
So where do I get a 2600 or 3600 with 16 ethernet ports for under $3000? -- -- *-----------------------------* Phil Howard KA9WGN * -- -- | Inturnet, Inc. | Director of Internet Services | -- -- | Business Internet Solutions | eng at intur.net | -- -- *-----------------------------* philh at intur.net * --
Anybody else have any problems with exodus, or is it just me? traceroute to beta.sierra.com (209.67.71.106), 30 hops max, 40 byte packets 1 gw.netmdc.com (206.29.138.1) 3.689 ms 1.867 ms 1.714 ms 2 netmdc-gw.abq-rtr-01.ihighway.net (206.29.131.21) 14.854 ms 20.363 ms 14.58 ms 3 bordercore2-hssi5-0-6.Bloomington.cw.net (166.48.176.25) 35.153 ms 77.288 ms 35.456 ms 4 core5.SanFrancisco.cw.net (204.70.4.85) 110.107 ms 65.928 ms 78.949 ms 5 * pb-nap-OC3-1.exodus.net (198.32.128.25) 51.899 ms * 6 * * * 7 * scca-05-p1-0.core.exodus.net (209.1.169.41) 86.462 ms 140.852 ms 8 scca-25-p0-0.core.exodus.net (209.185.9.30) 82.383 ms * * 9 * sewa-01-p0-0-0.core.exodus.net (209.185.9.186) 145.048 ms * 10 * * 209.67.64.21 (209.67.64.21) 131.238 ms 11 * 209.67.68.102 (209.67.68.102) 122.367 ms 153.178 ms 12 * * 209.67.73.2 (209.67.73.2) 139.211 ms 13 beta.sierra.com (209.67.71.106) 88.127 ms * 130.032 ms ZD(ekool@ns1)D(506/ttyp0)D(04:07pm:10/29/98)D- @D($:/etc)D- ping -c 5 beta.sierra.com PING beta.sierra.com (209.67.71.106): 56 data bytes 64 bytes from 209.67.71.106: icmp_seq=0 ttl=114 time=130.6 ms 64 bytes from 209.67.71.106: icmp_seq=2 ttl=114 time=191.4 ms 64 bytes from 209.67.71.106: icmp_seq=3 ttl=114 time=246.8 ms 64 bytes from 209.67.71.106: icmp_seq=4 ttl=114 time=132.1 ms --- beta.sierra.com ping statistics --- 5 packets transmitted, 4 packets received, 20% packet loss round-trip min/avg/max = 130.6/175.2/246.8 ms _ __ _____ __ _________ ______________ /_______ ___ ____ /______ John Gonzalez/Net.Engineer __ __ \ __ \ __/_ __ `__ \/ __ /_ ___/ MDC Computers/netMDC! _ / / / `__/ /_ / / / / / / /_/ / / /__ (505)437-7600/fax-437-3052 /_/ /_/\___/\__/ /_/ /_/ /_/\__,_/ \___/ http://www.netmdc.com [---------------------------------------------[system info]-----------] 4:05pm up 18 days, 19:34, 5 users, load average: 0.00, 0.03, 0.00
At 16:09 -0700 10/29/98, John Gonzalez/netMDC admin wrote:
Anybody else have any problems with exodus, or is it just me?
traceroute to beta.sierra.com (209.67.71.106), 30 hops max, 40 byte packets [...] 4 core5.SanFrancisco.cw.net (204.70.4.85) 110.107 ms 65.928 ms 78.949 ms 5 * pb-nap-OC3-1.exodus.net (198.32.128.25) 51.899 ms * 6 * * * 7 * scca-05-p1-0.core.exodus.net (209.1.169.41) 86.462 ms 140.852 ms
We're seeing the same thing. We've been route-map-ing around it since 10/02/98. On that date Exodus claimed that "MCI [C&W] is oversubscribed to the PB-NAP." Jim Browne jbrowne@jbrowne.com "I wish journalism would return to its glory days, when principled incorruptible men like William Randolph Hearst and Joseph Pulitzer were in charge."- Vanessa Jackson, Piano Teacher, www.theonion.com
We have been having problems reaching hotmail and ebay for weeks now.. Our customers called them, and they said it was within our network because they couldn't see us (traceroute clearly showed congestion on their side though). We finally did get some info out of them: Exodus net admitted that they have a bandwidth problem to MCI, and their border routers are severely congested. They are currently in the process of increasing their bandwidth with C&W. Their estimate (in their own words) is approximately 40 days until the circuits are upgraded. So, it will be quite a while before this will be remedied apparently. Erica L Johansson Network Administrator ServiceCo/Road Runner On Thu, 29 Oct 1998, John Gonzalez/netMDC admin wrote:
Anybody else have any problems with exodus, or is it just me?
8 scca-25-p0-0.core.exodus.net (209.185.9.30) 82.383 ms * * 9 * sewa-01-p0-0-0.core.exodus.net (209.185.9.186) 145.048 ms * 10 * * 209.67.64.21 (209.67.64.21) 131.238 ms 11 * 209.67.68.102 (209.67.68.102) 122.367 ms 153.178 ms 12 * * 209.67.73.2 (209.67.73.2) 139.211 ms 13 beta.sierra.com (209.67.71.106) 88.127 ms * 130.032 ms
On Fri, 30 Oct 1998, Erica L Johansson wrote:
Exodus net admitted that they have a bandwidth problem to MCI, and their border routers are severely congested. They are currently in the process of increasing their bandwidth with C&W. Their estimate (in their own words) is approximately 40 days until the circuits are upgraded.
I'm not sure I follow. What circuits is Exodus going to upgrade? AFAIK, Exodus has no circuits directly to MCI^H^H^HCW (for private peering)... Rather, Exodus exchanges traffic with CW at the MAE/NAP's, and via their Savvis transit ATM VC's (one to the Savvis NYC PNAP at 67 Broad St., and another so some Savvis PNAP in LA). Some clarification would be greatly appreciated. Thanks, -asr
Well, when I called them, the guy that I spoke with seemed a little clueless and wanted to keep blaming the problem on MCI. I had done some traceroutes and had seen other areas using various backbones having problems also. When I was on the phone with the guy, when he still maintained it was a problem on our side, I simply asked him what their connectivity was like, or what their network looked like. He had to put me on hold for several minutes, came back, and then started saying something about "We are working on a contractual agreement with C&W, but it will be approximately 40 days before we work that out." I had another coworker call later on..and that is what they had told us that time. I honestly don't have any idea what their network looks like, nor who they have connectivity and/or peearing with. This is just the information they had given us. They didn't give us really anymore..and the person that gave us this bit, seemed a little more clueful about what was going on. I'm just glad that they are finally acknowledging that there is _some_ problem on _their_ side of things. Now...hopefully they will have it fixed in the 40 days they are maintaining. Erica L Johansson Network Administrator ServiceCo/Road Runner On Fri, 30 Oct 1998, Adam Rothschild wrote:
On Fri, 30 Oct 1998, Erica L Johansson wrote:
Exodus net admitted that they have a bandwidth problem to MCI, and their border routers are severely congested. They are currently in the process of increasing their bandwidth with C&W. Their estimate (in their own words) is approximately 40 days until the circuits are upgraded.
I'm not sure I follow. What circuits is Exodus going to upgrade?
AFAIK, Exodus has no circuits directly to MCI^H^H^HCW (for private peering)... Rather, Exodus exchanges traffic with CW at the MAE/NAP's, and via their Savvis transit ATM VC's (one to the Savvis NYC PNAP at 67 Broad St., and another so some Savvis PNAP in LA).
Some clarification would be greatly appreciated.
Thanks, -asr
On Thu, 29 Oct 1998, John Gonzalez/netMDC admin wrote:
Anybody else have any problems with exodus, or is it just me?
Seeing the same thing. Got the same story from Exodus that it is a C&W problem, but I also see problems on traces not from C&W. -Chris ========================================================== Chris Candreva -- chris@westnet.com -- (914) 967-7816 WestNet Internet Services of Westchester http://www.westnet.com/
On Thu, 29 Oct 1998 08:45:36 +0200 (EET), Matei Conovici <cmatei@roedu.net> said:
Matei> Of course a Linux/PC will never beat a cisco :-) but the cost Matei> is sometimes an order of magnitude lower for roughly the same Matei> performance. It is just this point which keeps me running a PC router. I cannot afford to buy a cisco with 4 100Mbit ethernet cards in it. But I can afford 100Mbit PC cards all day long. Even one quad card. I would love to use IOS for the access lists and other goodies. rob
*shit. You have no clues. Linux was better at networking than BSD even in 1.2.x days ...
IIRC Linux didn't have VLSM in the 1.2.x tree, I seem to remember hacking GateD for ages to make it work at University only to find it wasn't GateD that was broken. The code for Linux hadn't been written. I then moved to work with NetBSD. In my opinion, and I've been using PC's and Sparcs to route for over 4 years now, all the Unices have their pros and cons, I use alot of NetBSD at home, but I use BSD/OS at work because of the support. I did use NetBSD with a ATM interface and BGP4 a while back and it did actually work although I doubt it could have driven 155M. There is a limit however that one hits when you really need hardware that is dedicated to routing. I recently have spent alot of time with several different vendors and to be honest all the vendors have their problems. The trick is to identify which one can deal with them the quickest, and there are some that need to wake up in a big way. followups to comp.os.*.advocacy Regards, Neil.
All the limitations of the Linux/PC router are due to the PC hardware architecture. As seen on the list, people put 8 cards in the same PC. This exceeds the bus speed of a PC. Even a single 100Mbps NIC kills the PCI bus in most PCs should it run full speed. Also, you have to be very carefull with the NIC you choose.
PCs simply were not built for forwarding packets and fast I/O.
Of course a Linux/PC will never beat a cisco :-) but the cost is sometimes an order of magnitude lower for roughly the same performance.
-- Matei CONOVICI, cmatei@roedu.net
-- Neil J. McRae. Alive and Kicking. Domino: In the glow of the night. neil@DOMINO.ORG NetBSD/sparc: 100% SpF (Solaris protection Factor) Free the daemon in your <A HREF="http://www.NetBSD.ORG/">computer!</A>
On Sun, 1 Nov 1998, Neil J. McRae wrote:
*shit. You have no clues. Linux was better at networking than BSD even in 1.2.x days ...
IIRC Linux didn't have VLSM in the 1.2.x tree, I seem to remember hacking
That's absolutely wrong. ------------------------------------------------------------------ Jon Lewis <jlewis@fdt.net> | Spammers will be winnuked or Network Administrator | nestea'd...whatever it takes Florida Digital Turnpike | to get the job done. ______http://inorganic5.fdt.net/~jlewis/pgp for PGP public key____
At 08:45 PM 11/1/98 -0500, Jon Lewis wrote:
On Sun, 1 Nov 1998, Neil J. McRae wrote:
*shit. You have no clues. Linux was better at networking than BSD even in 1.2.x days ...
IIRC Linux didn't have VLSM in the 1.2.x tree, I seem to remember hacking
That's absolutely wrong.
Guys, I'm a Linux advocate and a Caldera VAR, but even *I* can see that this whole thread is just plain wrong. It doesn't belong on NANOG. Evenso, Torvalds would NEVER condone the pro-Linux advocacy statements made in this thread. I'm not sure that the BSD developers would condone the pro-BSD statements either. BTW, except for one lone WinNT server, we are a pure Linux server shop. ___________________________________________________ Roeland M.J. Meyer, ISOC (InterNIC RM993) e-mail: <mailto:rmeyer@mhsc.com>rmeyer@mhsc.com Internet phone: hawk.mhsc.com Personal web pages: <http://www.mhsc.com/~rmeyer>www.mhsc.com/~rmeyer Company web-site: <http://www.mhsc.com/>www.mhsc.com/ ___________________________________________ I bet the human brain is a kludge. -- Marvin Minsky
On Sun, 1 Nov 1998, Neil J. McRae wrote:
*shit. You have no clues. Linux was better at networking than BSD even in 1.2.x days ...
IIRC Linux didn't have VLSM in the 1.2.x tree, I seem to remember hacking
That's absolutely wrong.
Hmm, could well be it was a long time ago. -- Neil J. McRae. Alive and Kicking. Domino: In the glow of the night. neil@DOMINO.ORG NetBSD/sparc: 100% SpF (Solaris protection Factor) Free the daemon in your <A HREF="http://www.NetBSD.ORG/">computer!</A>
At 04:26 PM 10/28/98 -0500, Adam D. McKenna wrote:
So does linux.
Linux can be:
A packet forwarder (router) A packet filter (firewall) An IP masquerading packet filter (NAT firewall) Can run RIP, BGP, EGP, OSPF (via gated)
Maybe I'm being naiive here, but what does Cisco offer beyond this (besides the availability of higher performance)?
Appliance level reliability, like a toaster. Plug it in, turn it on, configure it once, forget it exists. That's why we moved our printers to a dedicated print-server (OSIram), from the Linux hosts. Actually, a cisco may actually double as a toaster, a foot-warmer at the least <grin>. BTW, I thank the list for the kind help offered with our Linux NIC routing problem. It's still not solved, but we've been otherwise $distracted$. I plan on implementing some of the suggested solutions this week. Again, the issue is fall-back if our switch goes out again. For this sort of usage, Linux routing is far cheaper than keeping a cisco laying around. If the switch goes off, the lights go out, and the party's over. Linux routing would keep minimal lights on so the party could continue.
-----Original Message----- From: Michael Shields <shields@crosslink.net> To: dirk@power.net <dirk@power.net> Cc: nanog@merit.edu <nanog@merit.edu>; list@inet-access.net <list@inet-access.net> Date: Wednesday, October 28, 1998 4:01 PM Subject: Re: Linux Router KIT
In article <19981028100541.00359@orlando.power.net>, dirk@power.net wrote:
Linux doesn't just kill Microsoft's NT and Solaris. It also eats Cisco for lunch.
This isn't true. IOS does a lot more than just get packets from interface A to interface B. (In terms of managability as well as functionality) -- Shields, CrossLink.
___________________________________________________ Roeland M.J. Meyer, ISOC (InterNIC RM993) e-mail: <mailto:rmeyer@mhsc.com>rmeyer@mhsc.com Internet phone: hawk.mhsc.com Personal web pages: <http://www.mhsc.com/~rmeyer>www.mhsc.com/~rmeyer Company web-site: <http://www.mhsc.com/>www.mhsc.com/ ___________________________________________ I bet the human brain is a kludge. -- Marvin Minsky
participants (17)
-
Adam D. McKenna -
Adam Rothschild -
alex＠nac.net -
Christopher X. Candreva -
Dan Hollis -
Erica L Johansson -
Jawaid Bazyar -
Jim Browne -
John Gonzalez/netMDC admin -
Jon Lewis -
Lon R. Stockton, Jr. -
Matei Conovici -
Neil J. McRae -
Phil Howard -
Rob Walker -
Roeland M.J. Meyer -
Stephen Stuart