Show 112 – SDN’s Potential As A Displacement Technology with Jeff Doyle & Bill Koss

Well-known author and independent consultant Jeff Doyle joins Bill Koss of networking startup Plexxi as first time guests on the Packet Pushers podcast. Ethan Banks hosts, and Greg Ferro plays the role of in-house opinion-maker, trend-setter, and fashion icon.

What We Discuss

  • Let’s define SDN (again), because the definition has been shifting around as different vendors get into the mix.
  • Is Cisco’s push towards network programmability changing how customers perceive SDN?
  • If SDN ends up as merely fancy change management or service provisioning, is that really disruptive?
  • Assuming MPLS is overly complex for a data center deployment, does that point to an Openflow use-case?
  • Why can’t the network function like the other technology in an IT department?
  • There’s many pieces that must be put into place before SDN can start disrupting the status quo.
  • The importance of analytics engines as arbiters between network infrastructure (here’s what I have) and applications (here’s what I need).
  • What’s the relevance of legacy protocols like BGP and OSPF if we’re trying to reinvent the network to be software defined?
  • Risk aversion has resulted in 20 years of network inertia & fragility. How does that change to open the door to SDN approaches?
  • Teetering mountains of complicated code has built the network of today, but you can push it over with simple human error. And that’s bad.
  • Is it reasonable to expect large networking companies to provide SDN leadership?
  • Centralized or distributed controllers? Which is a better fit for SDN and why?
  • If you look at spending, it’s not all about cloud providers. It’s about the enterprise. So why so much focus on the cloudy corner cases?
  • If a bold new software defined network design gains favor, what does that mean for Ethernet fabrics and related technologies such as TRILL, DCB, 802.1BR, and SPB?

Links

 

Ethan Banks
Ethan Banks, CCIE #20655, has been managing networks for higher ed, government, financials and high tech since 1995. Ethan co-hosts the Packet Pushers Podcast, which has seen over 2M downloads and reaches over 10K listeners. With whatever time is left, Ethan writes for fun & profit, studies for certifications, and enjoys science fiction. @ecbanks
Ethan Banks
Ethan Banks
  • http://packetpushers.net/author/ecbanks Ethan Banks

    Plexxi is still in stealth mode (somewhat), and although they are doing demos with potential customers, their site is indeed a little vague on purpose. Short version – they’re making a merchant-silicon based Ethernet switch coupled with a controller & API. They are focused on traffic flows within a data center between members of an “affinity group”. I wrote in more detail about what Plexxi’s doing in the link below, although there is a register wall to read the article.

    http://searchnetworking.techtarget.com/news/2240158937/With-Plexxis-SDN-strategy-why-bother-with-network-fabric

    They’ve got a whitepaper that talks in more detail about “affinity groups.”

    http://www.plexxi.com/wp-content/uploads/2012/06/Plexxi-White-Paper-Affinity-Driven-Networking.pdf

    • http://www.itcertificationmaster.com/ Mirek Burnejko

      Thanks so much Ethan.

  • Citizen Private

    Hands down the best bit of analysis on this SDN silliness. Nice job Ethan and team.

  • http://twitter.com/nkrypted Brandon Mangold

    Good podcast as usual. Two points made in the podcast I am not sold on: A) I felt like there was some trivializing of SDN as a unified management system. SDN has huge potential as an efficiency gain which you began discussing towards the end with flow based control and the ability for an analytic engine to interface with applications and feed information to a policy engine.

    One big SDN key for me is that there is unlimited buffer at the host level and if we can tap into the potential to bring applications and thus hosts into the picture we can start to talk about more than loss-less Ethernet in the data center, how about virtually loss-less end-to-end networks with 100% link utilization… today’s networks are designed to run at 50% or less. A lot of wasted investment.

    And B) I believe the comments about Cisco’s inability to innovate were overblown or mis-stated. They have missed the mark on a lot of acquisitions but they are still able to innovate in certain areas of focus, look at UCS for example. I think the bigger issue is the internal politics leads to direction indecision and lack of agility to adapt to unforeseen market forces.

7ads6x98y