Slashdot Mirror
Open-Source Router to Take on Cisco?
ickypick writes to tell us that CNN is running an article about the emergence of an OpenSource Router product, currently in Beta, that targets mid-size enterprise customers for about one-fifth the cost of current enterprise networking giants' hardware. From the article: "The machine runs on two Intel chips, but far more noteworthy is its software, known as XORP, or extensible open router platform. The versatile open-source application can direct data traffic for a giant corporation as easily as it can manage a home Wi-Fi network." The current release is available for download from Vyatta's web site."
Seems like everything is Open Source now. (No, I am not complaining, i am backing it)
.... You name it.
We have Routers, Firewalls, IDS/IPS's, OS's, Word Processors, Spreadsheets, Presenting software. Hell. I would love to see an experiment where an entire corporate network was made, entirely of Open Source products (except for the hardware of course). From Routers to firewalls to
That would be an interesting, and totally free network.
Also very complicated
For a router, its mostly in the hardware, if it can keep up with real-life data rates.
Software is secondary..
---- Booth was a patriot ----
This is good since I always wonder how many back doors are in Cisco routers for Law Inforcement purposes.
Cisco's biggest advantage is their support network. I have yet to ever have a client that didn't buy smartnet with any of their gear.
Granted, some of their "engineers" leave a lot to be desired, but still, PHB's like the warm fuzzy feeling.
Make money? This better be good hardware running good software, because otherwise people are just going to say "fsck it, nobody was ever fired for buying Cisco". Why? Because Cisco actually works.
Yes, OSS community, your adversary actually works this time. Beware.
So who do you call when the thing breaks?
With Cisco, I call the rep, and they have a replacement device in our datacenter within the hour, and we load up the config and get it fixed.
Doubt you'll get that kind of service here, and that's what you pay for with Cisco.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
Dont you mean FreeSCO?
and that runs on pc hardware, this appears to be on custom hardware that can actually do the job. Using pc hardware only works for a small business.. the bandwidth isnt there.
---- Booth was a patriot ----
It can turn my old AMD K5 machine into a top-end Cisco machine. Does anyone have a spare ISA network card?
Grep. Gimp. Kugar. Krita. Kexi. LaTex. Tcl. And now, the piece de resistance - xorp.
Why route when you can XORP!
Imagestream has been doing this for ~8 years now ... course they provide support and all the hardware but this is doable. After all a DS3 Imagestream Rebel is only a P3 Intel and 256mb upgrade. Still it is another step in proof that cisco is not the networking god PHB's think.
I'm told you are what you eat, does that mean I can be you by tomorrow with some A1?
This could be a hit, if the costs keep down, for the small-medium business and home broadband markets. But I have trouble seeing how this will take significant market share in the Enterprise except for perhaps edge or LAN devices. For one thing, you pay Cisco, Juniper, Foundry, whomever for wire-speed implementations (among other issues) that rely largely on the ASICs and the overarching hardware architecture, beyond just the OS.
6 2391 for more info.
For the home market, there are already open-source software solutions such as for the Linksys WRT54-series wireless router, which is itself based on the GPL. See http://www.wi-fiplanet.com/tutorials/print.php/35
Until someone funds an open-source chip foundry, these won't replace the core.
The largest impediment is not software, but hardware. The two benefits to a Cisco are that (A) there is someone who *will* fix your problem for a fee, and (B) You can buy an interface card for ANY network type out there.
As for (A), the same will likely become available for this if it isn't already.
(B) is a lot harder. When you get into odd network types and high-speed telco lines, it becomes a bit more difficult - it isn't as easy as just calling your Cisco salesmonkey and buying the card you need.
It should be noted, however, that adding a card to a Cisco isn't always painless. I've had to upgrade the OS - which involved upgrading both memory and flash - just to support another ETHERNET card. How many decades has Ethernet been around for, and they want an OS upgrade to support one? And only to support an additional card, the built-in ethernet worked just fine.
Right now, we're using a Linux router for ethernet routing within our data center, which it handles just fine. As soon as our Sangoma cards show up, it's also going to handle a T3 to our office as well - but only clearchannel, we can't split it between phone and data (as I'd like to do.)
A while back, I had a rather perverse thought. You can hook up a LOT of interfaces to a high-end Cisco, and most routed telecom isn't very high-bandwidth. A T3, at a measly 45 megabit, is still very small considering the throughput of today's hardware. An OC3, at 155 megabits, still isn't much. The perverse thought was that if someone would come up with T1 and T3 modules with integrated CSU/DSUs that connected via USB or firewire, you could stuff a machine chock-full of 4-port controller cards, and be able to hook up 20 or more interfaces very quickly, and easily. In theory, each USB controller card *should* be able to push the ~200 megabits without much trouble, and even a plain old 32/33 PCI bus could *almost* handle the 110 MB/s of all 20 lines at full-tilt. Realistically, however, I do know that USB has many deficiencies which entirely prevent it from fulfilling that task.
Oh, you're not stuck, you're just unable to let go of the onion rings.
This is all assuming I'm willing to go unsupported, of course.
1) By far the most important is what kind of interfaces can I get for it. Of course I can get ethernet but what about T1, DSL, SONET, etc. If all this does is route packets over ethernet, which I then need to plug in to another router to get to my WAN, that's not so useful. I'd say over 90% of the Cisco routers I see in business are for WAN connections. If you are going to have to buy those anyhow, then what's the point?
2) What kind of load can it handle? Having something that can do a gig is all well and good, but can it still do a gig with 20,000 clients generating 50,000+ connections? That's where many budget routers and firewalls fall flat. They do everything in software so they can do the traffic no problem, but it's the concurrency that kills them.
3) Does it support layer-3 switching? That's where you in effect route the first packet of a flow and switch the rest. Leads to much lower impact on the router, and lower pings. Can't do it going from one media to another, but for internal routing it's the way to go.
This is, as mentioned, not considering support. I mean it's all well and good to slap some NICs in a system, load an OS that can route traffic, and call it a router/firewall/whatever, but it's something else entirely to see that survive under a real load. We see that all the time on campus when we test new potential devices. They promise gig throughput, something I have no doubt they deliver, and less than we use, but they instantly crash when exposed to our network. Why? Well we have like 30,000-40,000 comptuers or so that generate hundreds of thousands of concurrent connections. They just aren't equipped to process that kind of load and they stop passing traffic. The Ciscos, however, that compose the entire core, edge, and distribution parts of the network, operate without problems.
This seems to be alittle late to be jumping into this market. Most of the big players are starting to switch over to multilayer switching. Software routers are only needed where you need to do something like NAT or firewalling.
If your big enough to need a routing protocol like BGP, your going to need some serious hardware. Software based routers running on off the self hardware are fine for 100mbit ethernet routing, but beyond 100mbit you need some specialized hardware.
I really don't see any advanage this system has over a linux router with the usual tools(zebra/quagga, ip, ifconfig, iptables, ebtables, etc...)
God, root, what is the difference?
The game has long since moved from just forwarding packets to providing intelligence in the network. Now companies want integrated security, voince, application intelligence and application (l5-L7) optimization, QOS, high availability, etc.. none of which you'll find in an open source router. This is why the networking companies stay in business. If companies wanted cheap packet forwarders, they would have bought linksys, 3com, huawiei, hp or any other me-too commodity router. They didn't and Cisco won.
--- RFC 1149 Compliant.
The number 1 problem with Xorp is that it supports only a tiny fraction of standard Internet routing protocols. They don't have the developers to support anything more than a bare-bones software router. If you're only going to use what they have, it's no big deal. (NOTE: I am only including actual common routing protocols, here. There are over 150 routing protocols defined and implemented by somebody, but few routers support more than 3% and only the Really Major Routers even pass the 10% mark.)
The number 2 problem is that it lets the native OS deal with all of the QoS. This means that Xorp isn't guaranteed to behave the same on different platforms. It's not a lethal problem and some (including the Xorp developers) consider it a major bonus. I'm not convinced it's a good thing, though. It makes multicasting very confusing.
The final problem is that Click will normally be run as a kernel module, but Xorp is in userspace. This means you've a LOT of context switching when running in such a mode. Because you want minimum latency, the overhead of pushing packets into userspace in the first place might not be efficient enough.
I believe Xorp to be a good product. It is also the ONLY software router that is (a) Open Source and (b) being maintained (Quagga, Zebra and MRT are all dead, and GateD was withdrawn). I don't know if the Xorp group want more core developers, but I desperately hope that third-party developers offer patches and modules for it to beef up the abilities.
(Linux is an important software router. NetBSD and OpenBSD could be, if the routing software was good enough. The three of them should have the low-to-medium router market totally sewn up in no time flat, in a very short timeframe. That won't happen, though, if there's not enough independent interest and support.)
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)
Eddie Kohler, whose PhD thesis at MIT was the Click modular router (which from what I understand turned into the "engine" behind XORP), is one of the principal designers and developers of XORP. They published a paper at NSDI last year, which you can read here (Warning: PDF). It states very clearly what the goal of XORP is, and how well it performs. Quite interesting.
Except that the 'uninformed masses' are not ciscos main market.. we arent talking about twinkees here... ( and i know of one case where a bakery chain went down hard, due to one mistake.. the 'general public' understood what happened, and the place was out of business in 6 months, after nearly 100 years of being in the business )
Most of Cisco's market undersands the technology and security ramifications, and i think they would drop cisco in a heartbeat if this were to happen. Or at least i would hope they would...
---- Booth was a patriot ----
There are companies releasing high-end networking products that are nothing more than a PC motherboard and their software. A while back, one of the load-balancer companies (I think it was f5, but I don't recall for sure) contracted with Tyan to build their motherboards, with 4 (or more ) gigE controllers, each on it's own PCI-X bus, and Tyan also sold the board to the public.
The main reason that Cisco doesn't use commodity PC parts in their low- to mid- end routers is that if people knew they were getting nothing more than a $4,000 PC for their $15,000, they'd be pretty pissed. Also, there would be that many more people trying to "crack" IOS to make it run on white-boxes, and that opens up a whole new line of revenue drain for Cisco. (Not that people don't obtain unlicensed copies for their Cisco hardware, though...)
Oh, you're not stuck, you're just unable to let go of the onion rings.
OpenBSD ships with its own RIP, BGP and OSPF daemons. Its BGP daemon is BY FAR better than xorp and quagga, and its BSD licensed of course. OpenBSD is already a fantastic software router, maybe you should try using it instead of ignorantly telling us what it "could be"?
To a point, I agree with you. I like hardware; it just works. Flash back to several years ago when WinModems were first introduced... Remember what a disaster they were, especially for anyone who didn't fit the anticipated M$-using profile? They were cheaper yes, but also lower quality, more proprietary, and OS-dependent when compared to hardware modems. It was not very long until anyone buying a modem had to shop around very carefully to avoid being stuck with this type of shit. Because I do not use any Microsoft software (but they make decent mice), this was my experience before broadband became available in my area.
I don't want to see this happen to routers. With the reliability/availability that is usually demanded of a router, and the fact that routers are typically only implemented by either a knowledgable user or a hired technician, I do not anticipate this will actually be a problem.
However, I have encountered your "oh well they usually learn" arrogance before. Hell, from time to time I might display this myself. You know, the idea that anyone who disagrees with you or who wants to use a different solution for their needs than what you would use could only be suffering from a lack of education and must not have any valid point. It's just a dismissal. Dismissal is a favorite tactic of otherwise logical, composed people who do not care to truly examine a particular issue and are not honest about this unwillingness upfront.
The main question your post raises for me is that there is an unstated assumption there that Cisco is absolutely dominating this market (which I do not dispute) and is therefore THE sensible choice (this is the part I find questionable). Support contracts, features, performance, blah blah blah... To me these are not the central issue because you can get your desired balance of these by shopping around. So, just explain this one thing to me - how is a majority Cisco marketshare good for anyone other than Cisco?
FYI, I agree that software routers cannot match the raw performance of dedicated specialty hardware, but I also agree that fire is hot and liquid water is wet. I get the impression that neither Xorp nor any other software router is going to be marketed to Fortune 100 companies in an attempt to directly compete with Cisco, but rather is intended for small to medium sized networks. How many mom-n-pop setups and local businesses ever turn into multimillion dollar enterprises? For this reason I do not consider the "they all migrate one day" statement to be the showstopper that you seem to believe it is.
It is a miracle that curiosity survives formal education. - Einstein
Switches and routing are different things, you can't really compare the two. And again, in their router module, if you implement any sort of ACL, are you still avoiding process-switching?
This used to be the case waaaayyyy long time ago (ok we're talking years not decades) but starting in Cisco's Cat5500 series they've started pushing the FIB (Forwarding Information Base) into hardware as much as possible... Update an ACL and the assocated FIB gets updated. It started off with the first packet of a flow gets processed switch (i.e. routed) and then the rest of the flow after that gets switched after that, now with Cat6500s with a current supervisor card and fabric enabled host cards it's not even that. ACLs (now VACLs) modify the FIBs directly and everything is directly switched, TTLs decremented as they pass through, counters incremented etc (aren't ASICs nice)... allowing the processor lazely handle the hum-drum work of responding to SNMP requests that dump information tables that would chock a small horse.
-- Ed Bugg --You have freedom of choice, but not of consequences.--
2004 Cisco backdoor
2006 Cisco backdoor