Welcome!

Open Source Authors: Yeshim Deniz, Pat Romanski, Elizabeth White, Liz McMillan, Jnan Dash

Blog Feed Post

Plexxi Pulse – Celebrating Valentine’s Day with Abstractions

One month from today, Plexxi’s Nils Swart and Derick Winkworth will participate in SDNCentral’s DemoFriday to illustrate application abstraction using OpenDaylight. We encourage you to register here and check out the demo on March 14 at 10 a.m. PDT.  In our video of the week, Dan Bachman discusses how the Plexxi switching solution makes use of a fabric control protocol and explains whether there are any changes to how this works in a pod architecture. Here is our video of the week and a few of my reads in the Plexxi Pulse – enjoy!

 

 

This week Jason Edelman expanded on a prior post on the Cisco Nexus 3000 and its built-in Python interpreter and discussed an example that integrates Python with the native Linux operating system. I like how this post provides examples of real-world automation. As SDN emerged, many industry analysts and networking personnel started looking at what it meant for the future, but there is a lot of work to be done before SDN is widespread and successful. There are immediate practical changes that can happen within current architectures. This type of work is often more approachable and it helps network engineers add new skills to their repertoire.

CRN contributor Ashish Dhawan evaluates the SDN trend from the CIO’s perspective. He says CIOs are debating whether betting on SDN will impact their operations and the bottom-line. Ashish notes that SDN brings networking into the age of the cloud and Web 2.0 companies like Facebook are already embracing it. I would add that on the service provider side, SDN should be broader than just protocols like OpenFlow. Some of the protocols that are available to providers go well beyond simply turning up new services more quickly and with more automation. Providers should look at Path Computation Element (PCE), BGP-TE, and ALTO, and tracking emerging ideas like I2RS. These help make bidirectional communication between applications and the network more possible. If this happens, it’s not just that services are turned up faster but also that new classes of capabilities will be available. The path to new revenue streams (long an issue on the SP side) is not just doing the same thing more cheaply.

Larry Dignan at ZDNet  evaluates Cisco’s latest products prior to the earnings announcement made this week. Larry says while Cisco is making efforts to stay relevant in the networking space, big changes take time. Analysts recently observed that the threat of white box or bare metal switching will impact CapEx. This will pressure Cisco and all networking vendors to keep prices in line, but there are two things to consider:

  • The longer term cost bogie needs to be OpEx. It is by far the major driver of cost and competition will be on merits of the solution. Automation frameworks will also become increasingly important.
  • Pricing is the primary differentiator when two solutions are equivalent. The question is whether Cisco can maintain premium pricing because they offer a catalog of features no one else supports. The real threat of SDN is that architectures are becoming simpler and the catalog will be less relevant over time. So what does Cisco add to their portfolio to maintain differentiation? If ACI is any indication, they will say that it is integration with other infrastructure to provide better performing and more application-centric solutions.

Network World’s Jim Duffy reviews Big Switch’s change of course toward the orchestration of physical and virtual networks. Now Big Switch is following Cisco’s strategy of application centric infrastructure. There appears to be three network OS companies now: Cumulus, Big Switch and Pica8. They all cite customers like Google and Facebook. It will be interesting to see if those companies choose a single software vendor or if they maintain a dual-vendor strategy. It seems difficult for all three to make it with the same customers. More technically-savvy and demanding customers like Google require vendors to quickly ramp up support staff, which may be difficult for these companies initially.

Steve Evans at ComputerWeekly wrote about Dell’s open-networking initiative that involves a decoupling of networking hardware and software. Customers will be able to choose between Dell’s proprietary OS and a Linux-based OS offered by Cumulus. I don’t really connect Dell running Cumulus software on their switches to SDN. Whether your OS and hardware are tightly integrated or distributed separately is related to packaging, not to SDN. Just because the word “software” shows up in both doesn’t make them inherently connected. SDN is about separating control for the purpose of intelligent decision making and to automate workflows. The reason adoption is lagging is because the industry continues to confuse people. Vendor marketing around SDN is going to inhibit making money on SDN.

Brian Prince at Network Computing  says security is the major topic missing from SDN discussions in the networking industry. The real change here is that coordination across industry groups today is largely manual and governed by process. When you take out the human bottleneck, the pace of change accelerates. This exposes the human interaction bottleneck. I would think that moving to more deployment automation (as with DevOps) would be a natural evolution to tighten the processes and provide a layer of validation. This of course does nothing to add security. It just makes the security that exists a bit more tightly linked and verifiable.

InformationWeek contributor Bill Kleyman provides an overview of SDN’s three biggest benefits, which include complete cloud abstraction, intelligent global connections, and near-flawless content delivery. I think SDN primarily addresses workflow automation. Because we manage networks through pinpoint control on a device-by-device basis, things like edge policy are very manual and extremely costly. SDN’s major objective is to provide a central point for management, effectively automating a lot of the manual parts of managing a network today. Beyond that, the controller can add intelligence because it has a global view of the network. This means that the actual behavior can be tuned to application requirements. SDN allows the network to optimize based on what applications need. This happens through abstraction, but not just cloud abstraction.

Forbes contributor Ben Kepes writes about OpenDaylight’s recent Hydrogen release and asks whether it has any substance, arguing “fascinating does not equate to adoption.” At Plexxi, we think the inaugural OpenDaylight release is notable for two reasons: it represents working code, and it was done in 10 months. Many of the standards bodies and consortiums produce recommendations. ODP is producing code and it’s a big deal. The best way to bring nascent technology to market is through experimentation. We cannot debate endlessly in professorial exchanges and hope to really learn and iterate. Second, it would take most startups 2-3 years to get from inception to version 1.0 of their product and OpenDaylight did it in 10 months. What does this mean for the long-term trajectory? I think ODP represents a bright spot in the industry, and one that will ultimately prove successful.

The post Plexxi Pulse – Celebrating Valentine’s Day with Abstractions appeared first on Plexxi.

Read the original blog entry...

More Stories By Michael Bushong

The best marketing efforts leverage deep technology understanding with a highly-approachable means of communicating. Plexxi's Vice President of Marketing Michael Bushong has acquired these skills having spent 12 years at Juniper Networks where he led product management, product strategy and product marketing organizations for Juniper's flagship operating system, Junos. Michael spent the last several years at Juniper leading their SDN efforts across both service provider and enterprise markets. Prior to Juniper, Michael spent time at database supplier Sybase, and ASIC design tool companies Synopsis and Magma Design Automation. Michael's undergraduate work at the University of California Berkeley in advanced fluid mechanics and heat transfer lend new meaning to the marketing phrase "This isn't rocket science."