Stories
Slash Boxes
Comments

SoylentNews is people

posted by LaminatorX on Monday March 17 2014, @08:04AM   Printer-friendly
from the series-of-tubes dept.

evilviper writes:

"According to industry analysts, the reduced sales of traditional switching/routing heavyweights, during this traditionally active time, is due to widespread corporate investments and trials of software defined networking (SDN) equipment, which promises to improve routing efficiency, network management, and dramatically reduce hardware costs. Industry heavyweights like Google, Amazon, Facebook, and others are already heavily invested in OpenFlow and SDN, but it seems to be taking hold on a much wider scale, and not just in ultra-massive data centers."

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 3, Funny) by c0lo on Monday March 17 2014, @09:13AM

    by c0lo (156) Subscriber Badge on Monday March 17 2014, @09:13AM (#17469) Journal
    A Linux box with beefy network gear can be configured as a router? Who'd have thunk of this?
    --
    https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
    • (Score: 3, Insightful) by SlimmPickens on Monday March 17 2014, @09:42AM

      by SlimmPickens (1056) on Monday March 17 2014, @09:42AM (#17478)

      You make it sound like they're all booting up Monowall. Openflow is a big deal, every major network vendor is falling over themselves to implement it even though it heavily undermines they're business models.

      • (Score: 4, Interesting) by c0lo on Monday March 17 2014, @10:12AM

        by c0lo (156) Subscriber Badge on Monday March 17 2014, @10:12AM (#17488) Journal

        Openflow is a big deal, every major network vendor is falling over themselves to implement it even though it heavily undermines they're business models.

        If the older business model is failing, it's a case of adapt or die.

        Now, guess what? It is exactly because the existence and the openness of Linux/BSD and availability of network gear these advances are possible outside the control of network gear vendors. For instance:

        1. remember the good old "Beowolf cluster"? To my knowledge, it was among the first to support the link aggregation [wikipedia.org]
        2. more recently, the Wifi/3G/bluetooth connectivity in smart phones started to spawn strange ideas in the twisted mind of some (how dare they innovate outside network vendors?).
          Like, how hard would it be [usenix.org] to arrange TCP streaming [wikipedia.org] over multiplepaths [multipath-tcp.org]?
          (I remember picking it up on the current green feta site about a year ago. Uh, if you can stand the sight of it, here it is [slashdot.org]: March 22 no less)
        --
        https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
        • (Score: 4, Informative) by SlimmPickens on Monday March 17 2014, @11:34AM

          by SlimmPickens (1056) on Monday March 17 2014, @11:34AM (#17514)

          It is exactly because the existence and the openness of Linux/BSD and availability of network gear these advances are possible outside the control of network gear vendors.

          I fully agree. However what's going on is a lot more significant than a linux machine with hardware accelerated switching, bonding and multipath. We're talking about potentially not even using ethernet and TCP/IP. It is the fundamental shift to programmatically defined networks (originally designed to support virtual infrastructure) with abstracted hardware in the datacentres of those big companies in TFS and all of the network vendors scrambling to get on board. And you're waffling on about bonding and multipath.

          • (Score: 2) by c0lo on Monday March 17 2014, @01:25PM

            by c0lo (156) Subscriber Badge on Monday March 17 2014, @01:25PM (#17559) Journal

            And you're waffling on about bonding and multipath.

            What the hell? I read your opinion, I expressed a complementary - not contradictory - point of view. You really feel like a pissing contest about who's opinion is better or more coherently expressed? On Internet [xkcd.com], no less???

            ...

            ...

            (oh, my humble apologies sir.
            Please allow me to express how happy I am for your corrective intervention...
            I mean... such a soaring spirit, raising well above bonding and multipath, seeing... nay, scratch that, groking is the term... the socioeconomic undercurrents that shake and profoundly shape the nowadays society to announce a future that the mediocre mind can't comprehend.
            I envy you... how profitable and fulfilling must be your time spent on posting on soylentnews.)

            --
            https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
            • (Score: 2) by SlimmPickens on Monday March 17 2014, @02:32PM

              by SlimmPickens (1056) on Monday March 17 2014, @02:32PM (#17611)

              What the hell? I read your opinion, I expressed a complementary - not contradictory - point of view.

              I'm sorry. I did feel you were attacking me for not supporting Linux. My bad.

              • (Score: 0) by Anonymous Coward on Monday March 17 2014, @03:16PM

                by Anonymous Coward on Monday March 17 2014, @03:16PM (#17637)

                The "now guess what?" part sounded kind of aggressive.

              • (Score: 2) by c0lo on Monday March 17 2014, @03:18PM

                by c0lo (156) Subscriber Badge on Monday March 17 2014, @03:18PM (#17640) Journal
                Accepted. Rereading the thread, I learned my lesson too - should avoid rhetorical question that admit a snarky interpretation (the "Guess what?" one) - while I did not intend it as an attack, I can't really fault anyone that would take it like so; my apologies for that.
                --
                https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
          • (Score: 1) by mvar on Monday March 17 2014, @07:33PM

            by mvar (2539) on Monday March 17 2014, @07:33PM (#17764)

            It is the fundamental shift to programmatically defined networks (originally designed to support virtual infrastructure) with abstracted hardware in the datacentres

            Care to give an example?

            IMHO the problem with SDN is fully described in the following article:

            http://blog.cimicorp.com/?p=1665 [cimicorp.com]

            • (Score: 2) by SlimmPickens on Tuesday March 18 2014, @04:51AM

              by SlimmPickens (1056) on Tuesday March 18 2014, @04:51AM (#17931)

              Care to give an example?

              I assume you're asking about the hardware [openflow.org] abstraction [openflow.org]. Think how much fancy mellanox hardware is found in datacenters.

              IMHO the problem with SDN is fully described in the following article [cimicorp.com]

              I'm not going to address all the points in the article however I point out that TFS itself is about how sales of traditional equipment is falling due to the widespread adoption of openflow. I think this [informationweek.com] does a good job of explaining what drives adoption:

              While no one in their right mind suggests completely replacing Ethernet and TCP/IP, anyone who's struggled to automate data center load management in today's virtualized data centers knows that current networking protocols present a challenge. For companies to make the most efficient use of their virtualized servers, they must move workloads around their data centers, but doing so implies moving network connectivity along with performance assurances, security, and monitoring requirements. Today, that's either impossible to do automatically, or the method for doing it is highly proprietary. And virtualization isn't the only challenge--as businesses add more applications to their networks, they need to address the unique needs of those apps at a policy level.

              So basically, SDN is here to stay because traditional networking doesn't handle the dynamic nature of todays datacentres. You can start playing with openflow right now because it's integrated into openqrm and probably all of the other provisioning magic sauce too.

              Also, I said above "We're talking about potentially not even using ethernet and TCP/IP." I want to clarify that I said that to try and explain what SDN is. While ethernet and to some extent IP are on the chopping block, the Linux TCP stack is going to be with us for a very long time.

              • (Score: 1) by mvar on Tuesday March 18 2014, @07:19PM

                by mvar (2539) on Tuesday March 18 2014, @07:19PM (#18212)

                I'm not familiar with mellanox hardware, i'll have to check it out. I was asking more for an actual example, but the links you provided will do. I have no doubt that SDN is here to stay,my problem lies with the way it is being pushed down our throats as the next big thing while the exact "nuts and bolts" of it remain a mystery for the vast majority of network professionals. Same thing happened with the "cloud" - last time i visited the openstack homepage it was a shitstorm of buzzwords and you'd have to dig for hours in order to find some technical document. As the article i posted says:

                We have trivialized SDN into being trivial. In an effort to make SDN news, to get URL clicks for publications and PR mentions for vendors, we’ve left every complicated issue out of our processes. We don’t write or talk about them, or even develop them. No application segmentation. No next-gen operations and automation. No new network services—we just use SDN to produce the same Level 2/3 services we had before it ever came along.

  • (Score: 2) by SlimmPickens on Tuesday March 18 2014, @02:02AM

    by SlimmPickens (1056) on Tuesday March 18 2014, @02:02AM (#17895)

    I think this software defined networking has major implications for security. Der Spiegel writes [spiegel.de]:

    In the case of Juniper, the name of this particular digital lock pick is "FEEDTROUGH." This malware burrows into Juniper firewalls and makes it possible to smuggle other NSA programs into mainframe computers. Thanks to FEEDTROUGH, these implants can, by design, even survive "across reboots and software upgrades." In this way, US government spies can secure themselves a permanent presence in computer networks. The catalog states that FEEDTROUGH "has been deployed on many target platforms."

    and

    A document viewed by SPIEGEL resembling a product catalog reveals that an NSA division called ANT has burrowed its way into nearly all the security architecture made by the major players in the industry -- including American global market leader Cisco and its Chinese competitor Huawei, but also producers of mass-market goods, such as US computer-maker Dell.

    So maybe SDN will allow networks to fragment like Linux distros, and combined with security technology and sophisticated traffic handling, SDN might change the game for this kind of malware and hopefully give us a powerful new weapon against spying

  • (Score: 2) by sjames on Tuesday March 18 2014, @03:43AM

    by sjames (2882) on Tuesday March 18 2014, @03:43AM (#17922) Journal

    I've looked and looked, and I can't seem to find anything in SDN or OpenFlow that hasn't been possible for many years through well planned vlans and/or mpls. Yes, a different way to specify it, but not really a new capability.

    Does it really bring nothing new to the table?

    • (Score: 2) by SlimmPickens on Wednesday March 19 2014, @09:54AM

      by SlimmPickens (1056) on Wednesday March 19 2014, @09:54AM (#18476)

      I can't seem to find anything in SDN or OpenFlow that hasn't been possible for many years through well planned vlans and/or mpls.

      When you put it like that there's not much difference, but even the lowliest openflow switch has those abilities and has to connect to central controller before it can even do anything. It's a new baseline that is the exact opposite of the old distributed hardware paradigm. The real story though, is that SDN lays a platform for rapid innovation. I think the next ten years will see networking change quite a bit.

      • (Score: 2) by sjames on Wednesday March 19 2014, @08:11PM

        by sjames (2882) on Wednesday March 19 2014, @08:11PM (#18682) Journal

        I'm a bit at a loss still. I have always considered networks to be freely definable through configuration, though I prefer to keep it decentralized.

        I can see value in finer control and a standardized configuration, certainly. I guess it's just that the marketing hype leads one to expect something more fundamental and revolutionary than that.