If you haven’t noticed, software-defined networking (SDN) is kind of our thing here at SDxCentral. We spend a lot of our time talking about it, writing about it, learning about it, and figuring out what to do with it.
This year was no different, but we decided you’ve heard enough from us for 2014 and might want to hear from others who might know better. (Don’t worry, we’ll be back in 2015.) We’ll let the experts — four analysts and one vendor representative — take it from here.
Brad Casemore, director of data center networking, IDC:
The SDN market has its best years of growth ahead of it, but 2014 was an important year. SDN went from being this relatively obscure technology used by massively scaled cloud providers, to something that is now within reach of many companies within the market. We think it is poised for significant growth in 2015, and we believe it will become a pretty substantive market of nearly $2.2 billion in 2015.
The overlay has been very popular in the data center, but interest in OpenFlow controllers and switches persists, too. Approaches to realizing SDN incorporating automation tools and other southbound protocols and northbound APIs are also in the mix. The hyperscale folks who moved to virtualize their infrastructure ran into the limitations of traditional networking, and now that large enterprises are progressing through their cloud journey, they are meeting similar challenges, though obviously their businesses, infrastructures, resources, and requirements are different from those of the hyperscale giants. Still, the linkage between cloud and SDN became clearer in 2014.
Dana Cooperson, research director, Analysys Mason:
SDN progress this year pushed back against what had seemed to be telecom industry single-minded focus on NFV. The industry moved from having to define and explain what SDN was and how it could be used to improve communications services and network operations, to more concrete discussions about use cases and PoCs [proofs-of-concept], from how the technology could help save money to how it could help CSPs [communications service providers] make money.
One area of progress that was a highlight for me was Transport SDN. I’ve been an observer of the OIF‘s implementation agreements and interoperability demonstrations for many years. The T-SDN demonstration that member vendors and carriers completed this fall in conjunction with the Open Networking Foundation (ONF) was a critical milestone in industry progress. The demos tested transport extensions to OpenFlow as a southbound interface, vendor northbound interfaces to the SDN controller (and NMS), and specific service use cases (e.g., bandwidth on demand and cloudbursting between data centers).
We’re expecting 46 percent growth in SDN-related spending (software, hardware, and services) by CSPs in 2015 compared with 2014, and 56 percent CAGR from our 2013 baseline through 2018 to $1.83 billion.
Manish Rai, vice president of marketing, Meru Networks:
2014, looking back, is the year when we saw the potential of using SDN in the enterprise. SDN has lots of use cases in the data center, but there were no compelling use cases prior to this year that were articulated for campus networks. You have students looking to connect devices to WiFi, so we see dorm rooms as a hotbed of SDN adoption.
SDN started as a technology on the wired networking side. Wireless networking has embraced it in 2014, so another trend we would like to see is wireless-specific extensions to support more of the wireless vendors in 2015.
Peter Christy, research director of networking, 451 Research:
The trends that I thought were most important are:
- The adequacy of merchant silicon switch chips. None of these mega-scale networks depend on proprietary chips.
- Increasingly, these networks seem to use flatter, denser interconnection than hierarchical networks (leaf/spine) and for that reason depend less on complex, high-end chassis switches.
- The solutions emphasize configuration automation characteristic of DevOps models rather than sophisticated centralized controllers.
- The solutions look more like server software provisioning than the kind of SDN controller many have envisioned.
SDN is going to have a lot of traction in 2014 what with server virtualization, hybrid cloud networking solutions, and SD-WANs. I don’t anticipate much in the replacement of legacy networks with OpenFlow technology, in fact, I think OpenFlow isn’t where the most interesting action is. I think the networking industry is still in trouble. There continues to be brutal switch price competition (fueled in part I’m sure by the success of the merchant chips), and if companies like Cumulus or Big Switch are at all representative, a software vendor’s revenue will be smaller than if a legacy provider had sold the solution. I think SDN will enable all sorts of important system solutions, but [vendors will] not reap greater revenues as a result.
Stu Miniman, principal research contributor, Wikibon:
In general, we tend to underestimate the barriers and impediments to adoption of new technologies and in 2014, SDN faced strong headwinds. Networking is only behind the actual data center for longevity of life cycle, and this is a factor in limiting large-budget deployments or significant changes. While Cisco‘s ACI progress was one of the biggest SDN announcements in 2014, overall Cisco’s field is trying to delay consideration of new software-led technologies that might lead more to consider startups, white box switching or VMware NSX, which was often quoted as enemy No. 1.
Service providers and web companies are leading in adoption of SDN, so the general adoption of cloud and hosted private clouds will accelerate the movement of workloads to SDN environments in 2015 and beyond. Users have been asking the ability to acquire networking services via software instead of monolithic hardware-based solutions and the increased importance of open source ecosystems including Docker, OpenStack, and OpenDaylight are driving to enable this new networking economy.