Cartesian examines the disruption and opportunities created by software-defined networking (SDN). We look into the facts (not the hype) to emphasize what actually is happening on the ground. This includes our view of the SDN market size and its composition.
By Shahed Mazumder
Networks, Disrupted: Software-Defined Networking (SDN)
On November 3, 2015, at the NXTWORK 2015 Customer Summit, Juniper Networks unveiled its decision to embrace the white box ethos - breaking the nexus between Juniper hardware and Junos software. This disaggregation is specially targeted towards enterprises looking to software-defined networking (SDN) with growing interest and expectations. By allowing Junos to be installed on Open Network Install Environment-compliant 3rd party switches, Juniper essentially recognized that some customers will prefer other vendors’ white boxes to Juniper hardware to run Junos. This is a very powerful testament of the current momentum behind SDN. Juniper is certainly not alone in the middle of this disruption. Their bigger rival and industry leader, Cisco Systems recently added Embrane to their growing list of SDN related acquisitions – vCider, Cloupia and most notably Insieme Networks, to name a few.
Although SDN has lived up to its potential as “the future of networking”, so far it has been in an early adoption phase. But the promise of flexibility to direct on-demand network resources in an agile and intelligent manner is inspiring enterprises and also carriers, to have a more serious look at SDN. This increased level of interest is driving incumbent networking vendors to offer SDN choices. There also is a growing number of SDN-centric startups, many of which are being acquired by their larger rivals.
A Hybrid Approach
In a recent survey of enterprise CIOs by Morgan Stanley, 60% stated that they plan to deploy SDN within the next five years, with 15% planning to deploy SDN within the next year. In another survey of enterprise network professionals by Webtorials, 64% of the respondents mentioned that their focus area for SDN deployment are data centers, whereas 1/4th of the respondents cited inter-site Wide Area Networks (WAN) or campuses as their focus areas. Figure 1 shows that network professionals believe their network will be mostly hybrid based three years from now.
Figure 1: SDN Deployment Focus Areas
Source: Webtorials, 2015
This reflects a broad acknowledgement that SDN deployments will be hybrid – a mix of physical and virtual networks. This can be attributed to network users having a massive sunk investment in traditional technologies that they want to leverage. This also emphasizes that not everything can be virtualized. The complex and high speed core routing functionalities that Cisco/Juniper implements with purpose-built silicon in Application Specific Integrated Circuits (ASIC) are difficult to replicate using only SDN.
As SDN gradually makes its way outside of data centers, open communities such as Open Platforms for NFV (OPNFV) and ETSI-NFV have begun applying the basic idea of centralized programmability to WAN, core network and Radio Access Network (RAN). SDN is often cited together with NFV, which, after being introduced by 18 carriers at Mobile World Congress 2012, was initially positioned as a technology that would benefit, but not necessarily depend on SDN. However, the overlap between SDN and NFV has grown over time, with OpenSDN being driven by NFV use cases that take OpenFlow beyond the data centers and into carrier networks.
A Billion Dollar Market and Growing
Driven by increasing demand in the data centers, SDN revenue for the enterprise segment has produced triple digit growth over the past three years increasing from almost nothing (0.3% of overall enterprise networking spend in 2012) to a force to reckon with (7.1% of overall enterprise networking spend in 2015). According to IDC (Figure 2), enterprise SDN will be a $867M market in 2015, growing 126% from 2014. In 2016, this market is expected to surpass the billion dollar mark. Growth rates will decline as the market matures further, however by 2018, enterprise SDN revenue is expected to comprise 23% of overall enterprise networking spend with revenues of $3.2B.
Figure 2: Enterprise Networking & SDN Revenue ($M), 2012-18
Source: IDC, 2015
Interestingly, although SDN will gradually take over market share from traditional networking, the hardware (i.e. physical network) will continue to be the biggest component of the mix. IDC’s 2014 report (Figure 3) shows that SDN applications and the virtualization layer are expected to comprise only 31% of Enterprise SDN market revenue by 2018.
Figure 3: Enterprise SDN Revenue Composition by Segment, 2018
Source: IDC, 2014
This highlights that incumbent networking vendors will continue to be relevant in a SDN proliferated world. Dramatic cost savings by replacing hardware from incumbents with Commercial Off-The-Shelf (COTS) products will not materialize so easily. The value of networking know-how, expanded channel/partner network and, most importantly, operational learning curves will work in favor of incumbent vendors.
The SDN market for carriers, although smaller in comparison to enterprise market today, is expected to grow dramatically within the next 3-5 years due to an acceleration in NFV adoption. According to IHS Infonetics, The global market for carrier SDN software, hardware, and services is expected to grow from $103M in 2014 to $5.7B in 2019.
SDN Momentum to Continue
The challenges associated with SDN adoption are not necessarily technical. The key challenge lies in SDN shifting the bulk of the ownership of developing and implementing network functions from network vendors to network users/providers - enterprises and carriers alike. This is precisely what makes the SDN value proposition less clear today as it brings in new risks and responsibilities to the network users/providers. However, enterprises and carriers are aware of the requirements of organizational shifts and capability development. Since the potential benefits of SDN can outweigh the costs, SDN will continue to gain more ground to become mainstream. Hyper-scale network users have already demonstrated how they benefit from SDN adoption. Following their lead and support (e.g. Facebook led Open Compute Project), we are heading towards an ecosystem where others also can take advantage of this massive paradigm shift.