Finding true north for networking research
The past, present and future of networking research has been the focus of debate for an international panel of experts brought together by the 11th Annual IMDEA Networks Workshop*,
It was recognized that with at least 40 years of networking research already in the past, the field has entered in proper maturity, with all that is good and bad about it. Networks, and the Internet in particular, have become a critical infrastructure for society. They have become as important as the electrical grid, roads, highways, and other transportation systems. Maintaining and improving the Internet, therefore, requires a colossal effort, and this guarantees the continuing importance of the networking profession, especially its engineering flavor, for many years to come.
At the same time, however, innovation and research on networks, cannot neglect the particular networks that we have created and invested upon. This limits the scope of new ideas, designs, and redesigns that can be tested to those that are somewhat compatible with the specific characteristics, protocols, and interfaces, of the particular networks that we use and rely upon. In effect, as with the power grid and its standardized voltage and power outlet designs, the Internet has become too large to be changed or scrapped altogether, at least in what it is its main interfaces like IP, DNS, BGP, and a few others. This indeed limits the scope, and one may say even the importance of designing and proposing alternative network architectures, but then this is just a small facet of what networking research is all about. Important research remains to be done along multiple dimensions.
First, even with standardized interfaces, new technologies and trends such as Machine Learning, Software Defined Networks, and Blockchains, are revolutionizing the way that mechanisms and protocols are implemented behind those interfaces.
Second, the above mentioned interfaces are effectively found in what is often called the ``thin waist’’ of the Internet protocol stack. Whereas those interfaces are indeed difficult to change, the physical and application layer of the Internet, at the bottom and the top of the stack, respectively, are areas of vibrant new research and innovation.
Third, the omnipresence of networks, the vast array of technology vendors, customers, and applications to be supported makes operating such networks (OPEX) more expensive than designing and building them (CAPEX). Indeed, much of OPEX includes human labor, energy, real estate and spectrum costs that keep getting more expensive when, at the same time, CAPEX spent in networking devices keeps getting smaller, as a consequence of Moore’s and other technology scaling laws. Most of what we consider to be network research has gone into CAPEX related matters, leaving a huge amount of work to be done on OPEX optimization, or in other words, on operating more efficiently the networks that we already have. The latter, includes the business layer of the Internet, ranging from designing appropriate offerings and tariffs, negotiating deals with partners, and moving into new businesses such as content, home automation, or services around the Internet of Things and the Cloud.
Forth, networks are no longer only about carrying bits; those bits can be ideas, personal and sensitive data, important secrets, etc, and as such, can have profound impacts on social, political, and economic life. Issues of privacy, security, governability, information and disinformation spread open up ample new fields for networking research, be it over existing and standardised, or new networking interfaces and architectures.
Fifth, like in other mature fields, the experience and tools developed in the realm of networking research can lead to major breakthroughs in other adjacent, or not so adjacent, fields. Indeed, as mathematics revolutionized finance in the 60’s and the 70’s, networking research with its strong empirical and measurement based culture, its lack of fear for complexity and big data, and its tolerance to faults and uncertainty, can have a profound impact on diverse fields from computational social science, to smart city and autonomous vehicles, and even more distant ones, such as in the study of bias and discrimination and their impacts on social and political life.
For all the above reasons, networking research, be it on core telecommunication matters, or on interconnecting complex things in general, apart from having a proud past, also has a bright future ahead of it.