Software-defined networking was the key topic at last week’s Interop show. The topic not only dominated the keynote presentations but was also apparent in the vast majority of booths on the show floor.

The basic concept of SDN is to separate the control plane and the data planes of a network so that software can control network connections, potentially making things like network virtualization and direct application control of a network much easier. But SDN is being used in so many places that the term seems to be nearing irrelevance as so many network products suddenly get placed under an SDN moniker.

The prominence of SDN was notable in the keynotes. HP General Manager of Networking Bethany Mayer emphasized how the company’s new products enable a unified physical and virtual data center fabric, including a core router optimized to support data center virtualization and new switches. The company has already shipped 20 million OpenFlow ports. More interesting, Mayer said HP now has five SDN-based applications: supporting a virtual cloud network, sentinel security, load balancing, WAN bursting, and unified communications for Microsoft Lync.

She brought up a customer, Conrad Menezes, vice president of network and information security at Sears Holding, to talk about building an SDN network. He advised to keep it simple, don’t play the numbers game, do your own testing, and build to scale with long-term investment protection.

Juniper Networks’ EVP of Software Solutions Division Bob Muglia showed the company’s Junos V Contrail Controller, an open-standards controller for SDN networks. This is an overlay controller, which is meant to work with a new services platform and app engine as part of the company’s plans to split network into multiple planes. Muglia said 2012 was a year of SDN hype, but 2013 is a year of execution. SDN will demand business models transform to meet the increased importance of software and the coming evolution of networking, he said.

Cisco Senior Vice President & General Manager of Enterprise Networking Group Robert Soderbery used the SDN terminology least in this group, focusing instead of how “the Internet of Things” with mobile networks, pervasive sensors, and telemetry will lead to new categories of applications. He brought up NBA All-Star Kyrie Irving to demonstrate a basketball tracking app that could be used by the NBA.

Soderbery also talked about the companies’ Cisco ONE architecture, which shares the basic SDN concept of using a control plane to manage the network, in this case including existing Cisco hardware. He mentioned that MGM Resorts has created a mobile app that uses a guest’s location inside the hotel to provide directions to particular locations and special offers.

Another keynote included a panel focused on SDN. Moderator Eric Hanselman of 451 Research noted that while SDN is often seen as being a future technology, in reality a lot of the underpinnings of SDN are already here.

In this panel almost everyone agreed that SDN’s key improvements include automation and speed of provisioning. Martin Casado, chief architect for networking at VMware (and one of the founders of early SDN vendor Nicera), said the primary value of doing networking in software is automation. “You need to do business quicker,” He said, and speed and agility are key.

Rajiv Ramaswami, EVP and GM of infrastructure and networking group at Broadcom, said that every layer of the network is changing, leading to “flatter and faster networks,” with more “east/west” (application to application) connectivity.

Rajeev Nagar, a group program manager on the Windows Core Networking Team at Microsoft, said that in addition to traditional network vendors, applications are now becoming part of the network. He noted that some applications don’t need to know about the infrastructure, but some need to have tight interaction. For example, he explained that Microsoft Lync needs to know about latency in the network, while real-time gaming can really benefit from directly addressing the physical networking layers.

Another topic all the panelists seemed to agree on was that new silicon enhances SDN capabilities. Casado said the supply chain is changing with customers now talking directly to silicon providers. “A lot of SDN is about who the new voices are and how they influence the SDN decision,” he said. Indeed, most of the network processing vendors were showing SDN directions at the show, including Broadcom, Cavium, and Freescale.

All of this will change the role of network administrators and Ramaswami discussed how SDN is leading to the blurring of roles between network and server administrators. Nagar said Microsoft has “deployed what you could call SDN at scale.” Administrators still need deep visibility into the network, he said, and the capability to specify constraints and make changes, even as the system remains relatively autonomic.

Of course, SDN was all over the show floor as well. Enterasys was promoting its “one fabric” approach, which it said has been pushing the idea of a flow-based architecture since the mid-90s. The company emphasized that fabric for control and management, a unified wire and wireless network, and its CoreFlow silicon.

But the company did recently open up new APIs to integrate with third parties. This is a “northbound API,” meaning it works with applications as opposed to hardware-based APIs (such as Open Flow).

And it didn’t end there. Ixia, which is probably best known for layer 2 and 3 testing gear, is now pushing higher levels of testing for virtual networking. Arista showed off its new high-speed end networking switch, the 7500E, under a banner that read “software-defined cloud networking.” F5 Networks, which has been known for its load balancing and traffic management appliances, seems to have rebranded the concept as “application-level SDN.”

In some way, all this re-definition of everything as “software-defined networking” makes sense, in that applications and software are controlling network flows. But in other ways, it just means that almost everything in network can be considered SDN, which seems to water down the concept.

Source:http://forwardthinking.pcmag.com/none/311483-software-defined-networking-takes-center-stage-at-interop

Software Defined Networking is a term, a term that should you conduct a web search for it, will generate in excess of 5,000,000 results in a basic search; a few months from now, it would not be unreasonable to expect the number of results to have a zero, or a couple of zeros added to that number; such is the growth of a term.

For me, over a decade ago, I remember visiting with several of the Founders of the Internet, listening to IP and TCP began over 40 years ago; listening about for computational and other reasons, the limited design of what we still essentially use today came into being. It was easy to understand why routers were stateless forwarding systems, without visibility into application flows, and why there were no cross layer mappings or interactions.

Then as now, there was an understanding by our IT Founding Fathers of the limitations of the design and the need for network visibility into application flows and the need for statefulness, affirmative service delivery measures, and application level granular management of network infrastructure, in other words, the need for SDN-type capabilities, and they understood this need way before those who claim that 6, 7 years ago, they birthed the concept. Those and other members of that group of founders don’t get the credit for the implicit concepts that their work originated. They didn’t have the tools available today, and some have gone on to work in tangential directions, but there is no question that they, having invented all we use today, knew and still know more than any since.

To be clear, the value of SDN, is immense; it holds greater potential for productivity increases from IT than any development ever, simply because it brings together disparate elements; computing, networking, virtualization, information, and business logic. In physics, a Theory of Everything is a conceptual construct that links and explains everything; it is the intersection of all objects, conditions, interactions, state, and rules, both known and unknown. Essentially it is a unifying force between disparate elements.

What remains to be seen is whether SDN will be developed and implemented correctly, in short developed in such a manner that its true potential is realized and not derailed. To that end, the industry should be concerned with SDN that has lot of hype, but little production value; in Information Technology there have been many instances of over-promise/and under-delivery, which undermined the credibility and utility of the technological advance; at times when reading the claims of the most publicized SDN folks and large technology vendors, they make SDN feel like “cold-fusion”. A case in point is where two well publicized SDN companies recently accompanied a sole SDN “customer” in a discussion; the customer was referred to as a “unicorn”.

There are SDN customers, real customers paying real money, however they want real products. In short customers don’t want, and the industry doesn’t need a “movement”; they need real products for the real world pain points they deal with every day, and they don’t have the time, money, energy or inclination to do the custom engineering jobs that are being floated as SDN successes. Another case in point is a recent announcement at a trade show by the senior executive of a large technology vendor that they now had a “complete” SDN solution because they had hacked together an OpenFlow Controller, so now they had an OpenFlow Controller too; what was evident is that the executive did not have the faintest idea of what SDN is, why it is needed or its potential, nor do they even realize that their main competitor is using non-OpenFlow controllers to craft their own Cloud Strategy.

It would be interesting to question whether any of these executives even know that long before OpenFlow, 30 years ago, SNMP was one of the first efforts to manage infrastructure, but the granularity of CLI gained greater traction with administrators.

Proof that the industry should be concerned with obtuse executives at technology vendors technology can be found in this survey:

The results of a recent study of 250 businesses regarding SDN familiarity found:

• 48% replied “I have a general idea of what it is about”

• 35% replied “I understand the basics”

• 17% replied “I understand the details”

When more than four out of five potential customers do not understand SDN, one can reasonably conclude that this is due, at least in part, to poor messaging resulting from the paucity of understanding by executives responsible for the commercial development of SDN, and more ominously, that lack of understanding will almost certainly result in the mismanagement of SDN commercialization.

These two deficiencies will predictably hold the industry back. If there is any question of the size of the negative effects upon the industry, simply total the last $100 Billion or so of acquisitions that resulted in $100 Billion in write –offs, and decide for yourself whether the executives at these technology vendors and their actions are negatively impacting the industry, and then project for yourself what that holds for their management of SDN commercialization.

For the second time, a blog that I intended to write about technological issues; Northbound Interfaces, non-OpenFlow methodologies, and the over-arching direction of SDN, has taken a temporary backseat to the issue of how the industry is in danger of fumbling a technological advance that began with the IT Founding Fathers. I will write about those ideas separately.

Many may claim the concept of SDN for themselves, many want you to accept their claim so as to financially benefit them, and many have naively accepted their statements; however if you had been around when the IT Founding Fathers were discussing their ideas, you might think otherwise.

Source:http://www.wired.com/insights/2012/12/software-defined-networking-a-theory-of-everything/

Analyst Zeus Kerravala said Vyatta was not squarely in the software-defined networking camp but the rise of SDN gives Brocade a problem its technology can solve. “When you look at Brocade’s SDN offering, it’s limited primarily to LANs. Vyatta gives them a way to offer a software-based router to be able to connect those,” he said.

Brocade is buying privately held networking company Vyatta in an all-cash transaction. Financial terms of the acquisition were undisclosed. Vyatta created the industry’s first and only platform-independent network operating system, Vyatta Network OS. The OS uses software-defined networking (SDN ) to decouple routing and security software from proprietary hardware.
“This acquisition complements our R&D investments in Ethernet fabrics and SDN, as well as our broad industry and solutions-level partnerships that enable Brocade to pursue new market opportunities in data center virtualization , public cloud , enterprise virtual private cloud, and managed services,” said Mike Klayko, CEO of Brocade. “We are now bolstering these ‘build and partner’ efforts with this strategic acquisition with the goal of being the innovation and thought leader in the software networking category.”

Advancing Networks

Vyatta Network OS delivers advanced routing, security and VPN functionality for physical, virtual and cloud networking environments. The OS promises to empower system architects to secure and connect virtual, cloud or physical infrastructures while delivering the same elasticity, mobility, auto-provisioning and utility licensing benefits of the applications they support.

Brocade plans to leverage Vyatta technology to offer an end-to-end architecture built on a virtualized network infrastructure. This virtualized data center infrastructure aims to make both service providers and enterprises more agile by accelerating deployment and reducing development cycles in their networks.

Brocade said the technology extend its open data center fabric and IP networking technologies directly into any number of server hypervisor environments that gives customers the utmost choice and flexibility in selecting the right solution for their business needs.

“There are many significant developments happening today that are redefining data center architectures and industry landscapes,” said Ken Cheng, vice president of the Routing, Application Delivery and Software Networking Group at Brocade. “The Vyatta acquisition brings in considerably more software networking technology and expertise to Brocade.

“We believe software networking to be a critical component in the next phase of network virtualization as enterprises are becoming increasingly virtualized and actively moving workloads to the cloud.”

A Problem to Solve

Zeus Kerravala, principal analyst at ZK Research, said Vyatta was not squarely in the SDN camp but the rise of SDN gives the company a problem its technology can solve.

“When you look at Brocade’s SDN offering, it’s limited primarily to LANs. Vyatta gives them a way to offer a software-based router to be able to connect those. The technology allows them to make their SDNs broader,” Kerravala told us.

“From what I heard, they didn’t pay a ton for the company. Brocade acquired some good technology for a reasonable price and can drop it into their product set to augment what they do fairly nicely. There’s no overlap at all. There’s no conflict. And there’s no huge financial risk here.”

Source:http://www.sci-tech-today.com/news/Brocade-Buys-Vyatta-in-SDN-Play/story.xhtml?story_id=012001C8Q060