The SDN Revolution
Software Defined Networking (SDN) is one of the hottest new topics in the networking industry, as attested not only by all the headlines this architectural shift has gotten in the past year, but also by the intense activity within this market segment. Additionally, more than $400 million of venture capital was invested in SDN startups and over $1.5 billion in acquisitions related to the technology. Current trends such as agile development and the dev-ops movement are also contributing to a higher interest in SDN.
SDN represents a new paradigm within the networking industry, yielding more efficient, scalable, agile and dynamic networks, by increasing programming and automation. Its main benefits are reduced operational costs, lower capital expenditures (via improved resource utilization), faster provisioning, a marked improvement in the performance of networks and the promise of a more open, standards-based architecture, which enables a wider choice of suppliers for companies adopting SDN.
There are three key distinctive features in SDN architecture:
• Separation of the control plane from the data plane
• A centralized controller and view of the network
• Programmability of the network by external applications
The uptake of SDN will come from three market segments: enterprises, carriers and hyperscale datacenters (such as Amazon and Google).
The enterprise sector represents one of the low-hanging fruit for the technology, as firms seek to optimize their data centers. With the advent of SDN and Network Function Virtualization (NFV), the control software can now manage any switch from any vendor, in the same way as operating systems such as Linux and Microsoft can run on any generic PC or server. In an SDN topology, core capabilities such as networking and storage are decoupled from the compute server (typically an x86 server). This enables alternative architectural options for compute, networking, security, storage and other capabilities (such as switchable memory) to be flexibly hooked up via a standardized rack in the data center. All of these efficiencies allow for the design of more flexible and flatter networks, resulting in CAPEX savings of potentially 50 percent (in terms of networking costs).
Vertical segments such as healthcare and financial institutions such as the University of Pittsburgh Medical Center (UPMC), Credit Suisse and Goldman Sachs have been testing SDN in greenfield deployments. A broader uptake will most likely take place in 2015, once other use cases besides the data center take hold.
"With the advent of SDN and Network Function Virtualization (NFV), the control software can now manage any switch from any vendor, in the same way as operating systems"
The main goals driving the telecom operators’ adoption of SDN include CAPEX reduction, improved efficiency and the offering of new services. NFV can work hand-in-hand with SDN by virtualizing network services that run in dedicated appliances, such as Deep Packet Inspection (DPI), firewalls, load balancers and Session Border Controllers (SBCs), so that these services can run on a single pool of x86 computer hardware, yielding CAPEX and OPEX savings.
In order to accelerate the adoption of NFV, seven leading global carriers formed in January 2013 the Industry Specification Group (ISG) for NFV within the European Telecommunications Standards Institute (ETSI). The main goal of the NFV ISG is to formulate a standards based approach to SDN and NFV, in order to leverage volume and standardized x86 and switches to extend services on premise and in branch offices. NFV is complementary to SDN and vice-versa, while each technology is not inter-dependent with each other.
While some operators are actively seeking to deploy some NFV-type functionality, NFV maturity foreshadows a more prolonged cycle for adopting the technology. Carriers such as AT&T have stated that there are some areas they will not initially touch as part of their SDN/NFV initiatives, including the Radio Access Network (RAN), the routing core and the optical transport network.
This group is primarily comprised of Internet companies such as Google, Facebook, eBay, Amazon and Baidu, among others, which have been among the early adopters of SDN. Amazon and Google have embraced versions of OpenFlow, an open standards protocol to program switching or route hardware to forward packets (at either Layer 2 or Layer 3). OpenFlow is the most prominent variant to deploy SDN, as it is supported by over 70 members of the Open Networking Foundation (ONF), whose board members include Deutsche Telekom, Facebook, Google, Microsoft, NTT, Verizon and Yahoo.
Many of these early developments also entailed the development of home-grown operating systems (OS) by players such as Google. SDN will empower Internet companies implementing hyperscale datacenters to substantially reduce networking costs by using white box switches, controlled by separate control software. In the future, some of the players in this space might experiment with SDN controllers and third party operating systems from companies such as Big Switch, Cumulus, and Pica8.
In addition to these activities, Facebook and the Open Compute are further refining a specification for an open networking switch, which will be agnostic to the software running on it. Initiatives by organizations such as the Open Networking and Open Daylight Foundations are a big part of this effort.
The Final Word
Our outlook on SDN is not as bearish as that held by some Wall St. firms that SDN is likely to be near and longer-term negative for the incumbent networking vendors. Instead, we believe that SDN might actually offer a TAM expansion opportunity, with value shifting from hardware towards software and services. This will require a new, longer and more consultative sales paradigm, in which infrastructure vendors will need to refine their advisory services and become trusted partners by their enterprise and carrier customers. The lion’s share (circa 80-90 percent) of these vendors’ R&D spend is on software, as opposed to hardware, so it is only a matter of time before pricing models will reflect that.
Another change already happening is the advent of subscription pricing models, an alternative to the traditional perpetual license models requiring a steeper upfront CAPEX. With the subscription model, the Total Cost of Ownership (TCO) remains the same, but the upfront costs are reduced. That will increase the attractiveness of new SDN solutions to IT decision makers, but will create challenges to the Network Equipment Vendors (NEVs), as it will delay commission payments or channel compensation.
SDN will profoundly impact the telecom market, offering carriers and enterprises more choice, flexibility and increased network efficiency. Though the SDN market still is in its infancy, the technology represents a generational shift in networking that will span close to a decade, with a point of inflection occurring in 2015.