Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Networking Open Source IT Linux

Cumulus Releases GNU/Linux For Datacenter Routers 67

Posted by Unknown Lamer from the scale-up dept.
alphadogg writes "Start-up Cumulus Networks this week has emerged with a Linux network operating system designed for programmable data centers like the ones Google and Facebook are building. The company's Cumulus Linux OS operating system includes IPv4 and IPv6 routing, plus data center and network orchestration hooks. Much like OpenFlow for independent, software-defined control of network forwarding, Cumulus Linux is intended to run on commodity network hardware and bring Open Source extensibility to high capacity data centers. The head of the company used to work for Cisco and Google." The distribution is based on Debian and ported to several router platforms. They claim to release most of their code Open Source, but there are at least a few proprietary bits for interfacing to the routing hardware itself.
This discussion has been archived. No new comments can be posted.

Cumulus Releases GNU/Linux For Datacenter Routers More

Cumulus Releases GNU/Linux For Datacenter Routers

Comments Filter:

  • Can we finally replace Cisco now? (Score:5, Insightful)

    by erroneus ( 253617 ) on Wednesday June 19, 2013 @12:40PM (#44051203) Homepage

    What amazes me is the current level of brand name dominance in technology. (I know, I'm repeating myself so I'll abbreviate) It's all so consumerist. We don't care what it actually does or doesn't do... or even how well it does it. We just care about the brand name.

    "Do you know about networking?" "I'm Cisco Certified!!" "That wasn't the question..."

    How many Cisco certified people do you know that don't know anything about networking?? I know a lot. It's the brand name that makes them important and the brand name that makes these devices valuable.

    It tickles me to hear people say "Linux" and "toy" in the same sentence knowing that Cisco uses Linux in almost everything these days. That's like saying "I own a Lexus, you wouldn't catch me dead in a Toyota... those cars are crap!!" Sorry, but... you know?

    And to me the real killer is that networking is 99.999% about being protocol implementation faithful so ALL devices of all brands should do the same damned things. (Yes, I know there are Cisco specific protocols and people should avoid them to avoid vendor lock-in.) Cisco isn't quite as bad as Microsoft, but in some ways, they're worse.

    • Re:Can we finally replace Cisco now? (Score:5, Insightful)

      by evilviper ( 135110 ) on Wednesday June 19, 2013 @01:00PM (#44051449) Journal

      How many Cisco certified people do you know that don't know anything about networking??

      The CCNA was always an entry-level certification. Point me to a clueless CCNP or CCIE and I'll be surprised. Of course certification isn't a prereq for knowing what you're doing, either.

      It's the brand name that makes them important and the brand name that makes these devices valuable.

      There's definitely significant brand-name mark-up on their hardware, but that's NOT why Cisco is valuable. Cisco is dominant because of network effects... There are lots of people who know the Cisco ios CLI, and the intricacies of their specific product lines. This is in-part because Cisco is dominant, and it lends itself to more dominance, as saving a few hundred dollars on a switch isn't worth the retraining needed for your admin to learn to manage something different, or dealing with different support, billing, etc.

      And to me the real killer is that networking is 99.999% about being protocol implementation faithful so ALL devices of all brands should do the same damned things.

      Yes, and Windows and Linux can both serve web pages... Does that mean you'd just switch back and forth between them without a second thought? No, you'll stick with one or the other (whichever your IT staff is familiar with), until the benefits of the other become overwhelming, and are worth retraining or new hiring.

      Cisco hardware only needs to stay inexpensive enough that it's not worth the hassle of using a different brand. If you've got a huge network, that's not the case, and a few percent savings will add up. But for smaller networks, the savings usually really isn't worth the hassle.

      FWIW, I'm not a Cisco fan, but the business world isn't as irrational as you claim.

      • but that's NOT why Cisco is valuable. Cisco is dominant because of network effects

        This is the first time I see anyone arguing that market domination by brand X (in any market) is good because of the network effect.

        There are lots of people who know the Cisco ios CLI, and the intricacies of their specific product lines.

        Ditto here. The dominance of brand X is good because its products are so idiosyncratic that you need a lot of people to use it so that you could have a pool of people from which you could hire someone to manage it. I'm not sure what particular logical fallacy this is but I'm sure there is one.

        • This is the first time I see anyone arguing that market domination by brand X (in any market) is good because of the network effect.

          Since there's no anti-competitive, lock-in or monopoly aspect to Cisco's dominance, you'd be hard-pressed to claim that it is harmful. I specifically pointed out that the hurdle to switching to some other vendor isn't very high, just that a mix of different network equipment is anti-productive, even if there's some up-front savings to be had.

          The dominance of brand X is good be

    • Where I work, we're a pretty heavy cisco shop.... We've got 2 CCNA's (who are both going for ccnp) and our manager is CCNP Voice. We've worked on cisco gear for years, we have experience with it and understand how it works. Cisco has the market, thats a reason, for example, why HP CLI is extremely similar. So guys like me can QUICKLY go and configure vlans or something on a switch. Saving a couple hundred on a switch if your guys have to re-learn them isn't worth while.

      Would an OPENBSD rig running OOSPFD

      • That's ridiculous reasonig. It presumes people can't adapt their knowledge to new applications. I have no trouble with that.

        And Cisco people come more expensively than Cisco gear. Seems to me it would be cheaper to go with something "not cisco." This is why I said "maybe worse" [than Microsoft] because they REALLY leverage their critical mass hard and they do so by creating this army of Cisco loyalists because they all have the IT industry fooled into thinking they are worth more.

        None of what has been p

      • Re: (Score:2)

        by gl4ss ( 559668 )

        Where I work, we're a pretty heavy cisco shop.... We've got 2 CCNA's (who are both going for ccnp) and our manager is CCNP Voice. We've worked on cisco gear for years, we have experience with it and understand how it works. Cisco has the market, thats a reason, for example, why HP CLI is extremely similar. So guys like me can QUICKLY go and configure vlans or something on a switch. Saving a couple hundred on a switch if your guys have to re-learn them isn't worth while.

        Would an OPENBSD rig running OOSPFD or OBGPD be cheeper? YES. Would it likely be as fast? PROBABLY (if not faster). Would anybody know how to use it? PROBABLY NOT. While it could be learned.....time has a value. The exception is if you're starting from the ground up and have IT staff and procedures of size that could actually support 'non standard' gear. AKA if you get hit by a bus and you're the only guy in your shop -- how easy would it to be to get someone up to speed on your current gear? or do you need to fly in a specialist.

        TL;DR: We use the gear because we can find people who know it easily without having to fly in a specialist.

        sounds to me you guys should budget a day for training and move off cisco... if not for anything else as learn it as backup, if cisco decides to alter the deal since if you're willing to pay few hundred bucks more, why not a thousand or two thousand(actually depending on the gear you are already in that territory). basically what you just said is that you can't use anything else than cisco because you use cisco while if you used anyone else's gear you could use pretty much any of anyone else's gear. good go

    • Re: (Score:2)

      by dpilot ( 134227 )

      Coming at that from the other side, I've been running Linux somewhere between 15 and 20 years, certainly far enough back that there was nothing like a newby distro when I began. I have a small server cluster in my home, 2 subnets, etc. I've debugged some ugly problems at home and at work, where I'm one of the local experts. (as opposed to real experts, which I'll admit that I'm not.) Given the level of deprecation generally heaped on Certifications here and other sites in the culture, I've wondered how

    • Knowing Cisco and knowing networking are completely different things.

      I've known a few Electrical Engineers whom I've asked about tecnical information on IPv6 routing. They insisted I didn't need to know that much (I was building my own routing box), that if I used cisco just enabling a checkbox would help me do what I wanted, there was no point in doing thing myself.

      Of course, all of them cisco certified.

    • Cisco just opened up EIGRP if you haven't heard. [cisco.com] Maybe Cisco isn't so bad. It is not Cisco's fault that there are incompetent people with credentials. It happens in every field. Cisco does make some good products. From my experience the Cisco 6500 series devices last for eons. I know I probably sound like a fan boy. I agree that creating proprietary protocols is harmful to the industry and customers. I may be wrong and a bit idealistic, but I think that companies that make good products can make money even
    • Yeah, last time I checked Cisco supports cycles in VLAN's which is mathematically incompatible with graph theory which is what helps keep networking sane

      . Makes it incompatible with every other switch.

      . They are also about 2-3 times more expensive than quality, protocol following stuff like, um I guess Foundry is Brocade now or whatever. It has been a few years since I had to do some consulting for a research based MAN. For $30,0000 their Big Iron was on par with the $125000 solution a Cisco whore quote

  • Andreeson Horowitz funded startup (Score:4, Informative)

    by charnov ( 183495 ) on Wednesday June 19, 2013 @01:45PM (#44051959) Homepage Journal

    This is a Andreeson Horowitz funded startup founded in 2010. The principles are JR Rivers (formerly of Cisco and Google) http://www.linkedin.com/pub/jr-rivers/3/3b7/372 [linkedin.com] and Nolan Leake (formerly of Tile and 3Leaf) http://www.linkedin.com/in/nolan [linkedin.com]. They're pretty darn smart cookies.

    That said, I wonder if they are trying to gain some momentum (there seem to be quite a few major players in the SDN crowd since they founded their company) or if they have run out of steam and are trying to get the Open Source crowd involved on the development side... worth keeping an eye on, I guess.

  • Where to buy? (Score:3)

    by rainer_d ( 115765 ) on Wednesday June 19, 2013 @04:00PM (#44053463) Homepage
    Do they only sell to "large" customers, buying a couple of hundreds a time?
    What does such a switch cost, compared to a Switch from Cisco or HP?

Slashdot Top Deals

"If it ain't broke, don't fix it." - Bert Lantz

Close