Software-Defined Networking Promise Inches Closer To Reality

Now that Cisco has officially blessed SDN (possibly more like damned with faint praise, but that’s another story) and brought along a lot of influential friends to its coming-out party (including BFF/frenemy VMware, but again, we’ll save that for future follow-up), it seemed like a good time to update what’s happening in the software-defined networking universe. The short answer is: a lot!

Reportedly originally scheduled for October’s InterOp New York, Cisco’s $850 million SDN spin-in, Insieme Networks, made its first – and last – public appearance on Wednesday when the networking giant announced it was buying back the remaining 15% of Insieme it didn’t own, and unveiled the Application Centric Infrastructure. ACI is comprised of the Application Policy Infrastructure Controller (APIC), the Nexus 9000 portfolio, and enhanced versions of the NX-OS operating system. Companies that rallied around the ACI flag included BMC, CA Technologies, Citrix, EMC, Embrane, Emulex, F5, IBM, Microsoft, NetApp, Panduit, Puppet Labs, NIKSUN, OpsCode, Red Hat, SAP, Splunk, Symantec, VCE and VMware (of NSX fame).

The Cisco news comes after HP’s networking group announced an SDN developer kit and app store, and before next month’s scheduled OpenDaylight (the open-source SDN association whose membership includes Cisco and HP) code release. When asked if the Cisco news clarified or muddied the SDN waters, Charles King, Principal Analyst, Pund-IT, answered “yes”.

“Basically, ACI is Cisco’s alternative to the OpenFlow SDN framework developed by VMware/Nicira so it can be viewed as analogous to Microsoft’s Hyper-V. In fact, both ACI and Hype-V were created by vendors deeply concerned about the effects of VMware on their core markets which says something about the level of disruptive innovation VMware is delivering. It’s also interesting to note that (according to a Business Insider report) the Cisco Nexus 9508 switch that enables ACI contains two chips—a “standard” version that can be used in conjunction with other SDN products and a custom chip that delivers superior performance but only works with Cisco’s solution. So Cisco’s basic though unstated message is: Yes, we support SDN of every sort but if you want really optimal performance, you have to lock yourself into our proprietary SDN framework.”

Questions about just how open Cisco/ACI will be are apparently not troubling the who’s who list of vendors that have announced support. King said the “apparently counter-intuitive involvement of EMC, NetApp and even VMware simply reflects Cisco’s clout in enterprise networking (where the company reportedly owns about 70% of the market). At the end of the day, vendors who sell customers what they want to buy tend to succeed better and more consistently than those who try to strong arm clients into unwanted purchases.”

According to two new reports, the future for SDN looks very bright. The market is expected to grow more than six-fold over the next five years. The majority of the SDN market will comprise Ethernet Switches and Network Security Appliances, which are forecast to deliver 75% of total sales revenue in 2013.

“Almost every major Ethernet Switch vendor with exposure in the data center is announcing significant new products over the next several weeks” said Alan Weckel, Vice President at Dell’Oro Group. “To put perspective on the order of magnitude of the data center equipment market, in 2013 sales will exceed $100 B, split among Servers, Storage, Ethernet Switches, and Data Center Appliances, etc.”

As SDN gains momentum through 2H13 and into 2014, there will be additional opportunities for suppliers to sell more hardware; however, most customers will be unwilling to “rip and replace” their infrastructures with new hardware and will instead take a piecemeal approach to deploying SDN (e.g., deploying an SDN controller), according to the 1H13 Enterprise Network Vendor Benchmark from Technology Business Research. While hardware spending growth has been in low single digits, TBR expects that services, which accounted for 26.9% of total revenue for benchmarked suppliers in 1H13, will grow in 2H13 as suppliers such as Cisco, F5, Huawei and Siemens Enterprise Communications roll out new service offerings to their customer bases.

That’s all well and good for the big vendors – and their larger customers – but for the rest of the world, SDN remains a distant dream, according to Allied Telesis. In a recent interview with IT Trends & Analysis, the smallish (2012 revenues of $283.7M) networking vendor said there will be a lot of “storming and norming” before the SDN space shakes out.

“We’ve got a large customer base with real problems today,” said Mike Dunbar, President, North American Sales, Services, and Marketing. Having recently moved over to Allied from VMware, he does expect the network to get more commoditized, but that transition will take years.

A former member of the Marine Corps, where he was responsible for their networks, Dunbar remembers the manual effort and amount of time that went into making a router or switch change. “It was extremely complex and time-consuming.”

So yes, SDN makes sense, and it will take its place in the networking firmament, but in the meantime the company has created a bridging solution, the Allied Telesis Management Framework. “It’s a lightweight way to do SDN-like things in an AMF infrastructure,” he said.

Based on SDN principles, AMF is an embedded technology within the AlliedWare Plus OS that is native to Allied Telesis switches that simplifies the management of a network by automating and assigning everyday tasks, such as: making configuration changes to multiple units; backing up configurations; rolling out a firmware upgrade; adding new units to the network; and, recovering failed units with new units. “At one level, AMF technology is about automating and removing some of that complexity, said Dunbar.

There was a lot of debate at the company about if, when and how about going to market as SDN. He said it will be extremely relevant at some time, but AMF addresses today’s problems, and will eventually – two to five years from now – help customers make the transition to SDN when it becomes more market-ready. VMware doesn’t spend $1.2 billion for a kind of, sort of company (Nicira) that someday will be relevant, Dunbar added.

According to an InformationWeek 2013 SDN survey:

-33% of enterprises have no plans to test SDN and another 13% say it will be at least 12 months before they start;

-47% cite product immaturity as a barrier to adoption, including questions about standardization;

-44% don’t know what Open Daylight is; and,

-46% don’t buy in to the “dumbing down” of switches and routers with SDN.

Yet for more than half, 53%, automation and provisioning management was the top benefit sought.

So once again, the SDN debate comes down to when, not if. With the complexity, cost and the looming – and massive –  network obsolescence, industry and user associations and vendors like Cisco, VMware and even ‘small fish’ like Allied Telesis, will continue to bring that date closer.



Author: Steve Wexler

Share This Post On


  1. Interesting article.

    There continues to be confusion in how we talk about SDN. The AMF individual talks about SDN leading to commoditization and then in the next paragraph we are back to talking management and control.

    The real problem being solved is not a “HW is too expensive” problem. SDN is solving an edge policy management problem. The long-term bogie is OpEx.

    CapEx will come in line too, but not because of SDN. Prices drop when competition thrives. SDN is not forcing solutions to look identical. It is actually accentuating strategic differences. This should allow for even more price diversity. But competition (lots of new players in networking) will drive prices downward. Combine the competition with a convergence on a narrower set of switching silicon and we should see a better world for customers.

    -Mike Bushong (@mbushong)

    Post a Reply
  2. I thought one of the (main) benefits of SDN will be the ability to add new features and functions without having to do a rip n’ replace of your existing SDN network infrastructure. That should deliver both CapEx and OpEx savings.

    Post a Reply
  3. AMF lowers op-ex by reducing the amount of time your expensive, skilled staff spend on mundane, everyday tasks like configuration management, firmware upgrades and installing or replacing network devices. It doesn’t preclude SDN and actually facilitates rapid changes to network configs by making the whole process much easier and faster.

    Post a Reply

Leave a Reply