SDN: A Perspective of Protracted Debut, First Scores and Broadening Promise
It is an interesting aspect that the technology, applications and use cases related to Software Defined Networking (SDN) continue to evoke great interest as it would usually for a new concept that started a year or two ago. In fact, SDN was first introduced and demonstrated in small experimental networks as early as 2010 and the various key components of SDN were defined rather comprehensively at least to address Local Area Networks. It is noteworthy that SDN adoption could very well be one of longest in recent times—still emerging, evolving, and promising.
The hype cycle of SDN technology is definitely extended and skewed with the plateau of productivity still to reach—perhaps there is more than one peak of inflated expectations in the SDN hype cycle. After more than seven years of existence only a handful of SDN applications and use cases are currently deployed that could be monetized and justified with ROI. Still the excitement of SDN, its benefits and potential applications are ever high and on a good trajectory after going through multiple highs and lows. Technologies that usually take many years for adoption will be replaced and superseded by newer concepts and approaches. In this context one could recall how robust technologies such as ATM and WiMax faded away. Yet is it clear that SDN continued to slowly and gradually take foot hold, beat the odds and is here to stay because of its simplicity, transformative nature, and powerful applications.
While the equipment manufacturers were bringing products into the market, their customers, the Telco Operators had other priorities and were defining Network Virtualization Functions and network service architectures realized from the NFV components. The traditional customers were prioritizing and asking for a different set of technology solutions and their budgets clearly not aligned with SDN deployments. But SDN benefits are complementary and implicit that it eventually became a key component in the NFV architectures. This could perhaps provide some insights into this unique concept, which not only survived but also continue to broaden its use cases and show promise.
Today, hardly a week passes without a significant announcement of a collaboration between a CSP and a SD-WAN solution provider
SDN when it was introduced in 2010 had two main value propositions put forward—ability to run centralized control and user applications that can be monetized by the service provider and the ability to create virtual networks, which can be experimental in nature. By separating the data and control plane processing functions in a network node, one could move the control plane functions away from the network node, consolidate control functions of several network nodes, and manage the network end to end from a centralized controller. This elegant and yet simple architecture opened up powerful possibilities—ability to write SDN applications that have near real-time visibility and control of the network. Another related advantage is the ability to virtualize the network beyond the existing VLAN technologies and provide an end-to-end virtual network that could be used for experimental services before live deployment. An advantage that is perceived to be useful is reducing risks and improving efficiencies in service introduction.
However, initially network virtualization concept and its related use cases took a front seat in SDN deployments. Before user SDN applications that can be monetized are deployed one has to deploy SDN enabled networks. There is interdependency between them and it was becoming apparent that a near term ROI for application deployment was not possible.
But given the flexibility of SDN deployments, several equipment vendors started positioning their SDN platforms as not only providing the promise of future applications but also able to replace and manage current traditional network management systems.
SDN like its predecessor network technologies is evangelized by equipment manufacturers. But their customers— the telecom operators had a different set of priorities. In order to compete effectively in their markets, the operators were defining requirements for cloud based new generation of network service architectures and concepts around a technology called Network Functions Virtualization. Since 2012 much of the telecom operator’s focus was on defining NFV framework and components in detail for design and implementation. Several sub components of the NFV services architectures were identified and their requirements defined.
A good breakthrough came up here for SDN. Network Virtualization has become a key requirement within the NFV architectures to implement life cycle management functions of a cloud-based network service. An Open source SDN controller was adopted as a component in network services of OpenStack, a virtual infrastructure manager for telco NFV cloud. This renewed the excitement for SDN in the industry within both the vendor and operator communities. SDN has finally hit its first widely accepted use case with immediate priorities for deployment and its justification subsumes not only the benefits of SDN but also NFV.
SDN platforms were evolving since 2014 to support Life Cycle Management of cloud based components and services. There is also an alternate architectural approach where SDN can integrate with separate systems that provide life cycle management functions, replacing traditional network management systems called EMS and NMS systems.
Since 2015, another emerging application for NFV that is also propelling SDN further in deployment is SD-WAN. Virtualizing CPE functions and a software-defined Wide Area Network has provided a very good use case that is less complex to implement and with near term ROI for operators. With SD-WAN it is possible to provide remote service manageability, deployment, and new service introduction without expensive truck roll-outs and long delays in service introduction. SDN platform in this context is positioned as the platform managing end-to-end services in the SD-WAN network.
However, SDN benefits are independent of NFV and it is a technology that predates NFV. It has its own powerful set of deployments and use cases. While NFV provided much needed support, SDN continued to be adopted independently in other deployments. The power of SDN is network service agility, scale and programmability on demand.
It is likely more independent SDN deployments that are not necessarily intertwined with NFV will emerge. This will finally deliver on the promise of monetizing network control and user applications. Applications such as network monitoring, network security, network analytics and user trends, network optimizations and utilizations, data center interconnect and consolidations through network virtualization are few of the emerging ones. It is the technology that will deliver on much awaited and long discussed concept of end-to-end Quality of Service or User Experience. SDN will perhaps be the technology that will still be providing interesting applications and broadening its use cases much long after NFV has matured.