The Origin and Evolution of Open Programmable Networks and SDN

Telematika.ORG
@telematika_org

In this article, we review a number of early works on network programmability that illustrate how several features of SDN emerged progressively over several decades. Our review starts from the early concepts of network control in the telephone network and continues to examine a prolific period of research advancements in the 1990s and early 2000s that led to a number of start-up companies that followed. This also include IEEE’s own efforts in standardizing network programmability, and finally the arrival of the OpenFlow standard. We study the importance of this architectural transformation and its influences on modern cloud computing and next-generation networking.

The main concept behind SDN is the separation of network control and data transmission through open (and preferably standard) APIs (Application Programming Interfaces). Network operators are attracted to SDN because it provides an abstracted view of infrastructure that can break hard dependencies on proprietary vendor technologies and protocols and use commodity IT equipment to implement core functions such as firewalls, proxies, and other middle-boxes. SDN helps multi-tenant data centres where network services and connectivity are virtualized, and resources can be easily allocated to each tenant.

In the 1990s, academia was at the forefront of this movement. Still, the industry at large was not ready to adopt research projects from universities. Except for AT&T/Bell Labs (though one may argue that Bell Labs was also a research institution), telecommunication carriers were not supporting programmable networking as a design principle even though they were very interested in open architectures that would allow them to avoid the “single vendor trap”. On the other side, academia did not have enough leverage or resources to push for this novel concept to be accepted in the industry. Had one of the major network equipment vendors supported this effort at full throttle, SDN would have happened before 2000.

It took many years, if not decades, and a confluence of factors for the industry to drastically change its perception: the emergence and eventual dominance of Internet technologies and their decentralized model of operation, virtualization technology becoming dominant in data centres, the booming start-up ecosystem of the 1990s, the emergence of the API economy and application marketplaces, and a more favourable regulatory environment.

As these lines are written towards the end of 2020, multi cloud and hybrid cloud are becoming the norm for scalable computing. The Internet of Everything (IoE) and edge computing are increasing the complexity of networking. Other trends such as containers, microservices, and serverless have accentuated the networking properties of modern cloud applications.

In retrospect, all these efforts were incremental steps toward an architecture for the separation of control and data planes that would be acceptable and advantageous for equipment vendors, network operators, and network users. It took several decades and the persistence of the early researchers and the first entrepreneurial efforts to set the stage. And it was not until the emergence of OpenFlow circa 2005, when these ideas were finally able to reach critical mass and be widely accepted. That was the singular moment when manufacturers and network operators were finally ready to embrace an architecture based on open interfaces, and to overcome their prior reluctance to deploy open systems. Cost was also a motivating factor. Not only do open programmable networks offer better service creation opportunities, but they are also cheaper. They meet the expectations of network operators for performance, economy, and reliability, and of network users for capacity and QoS. And as new requirements emerge, these concepts maintain their relevance and prove to withstand the test of time.