Networking Research - A Reflection in the Middle Years

TL;DR

This paper reflects on the evolution of networking as a mature infrastructure, emphasizing the slow pace of change driven by standards, compatibility, and organizational factors, and discusses challenges for research impact.

Contribution

It offers a perspective on networking as civil engineering, highlighting the long timescales, standardization challenges, and industry dynamics affecting research and innovation.

Findings

Networking infrastructure has matured and is driven by standards and compatibility.

Research impact is limited by slow standardization and industry barriers.

Telecommunications operates like airlines, emphasizing marketing over innovation.

Abstract

Networking is no longer a new area of computer science and engineering -- it has matured as a discipline and the major infrastructure it supports, the Internet, is long past being primarily a research artifact. I believe that we should consider ourselves as the civil engineers of the Internet, primarily helping to understand and improve a vast and critical infrastructure. This implies that implementing changes takes decades, not conference cycles, and that implementation is largely driven by compatibility with existing infrastructure and considerations of cost effectiveness, where resources that research focuses on, such as bandwidth and compute cycles, often play a much smaller role than limited organizational capacity for change. Telecommunications carriers, in particular, have become akin to airlines, largely operating equipment designed by others, with emphasis on marketing, not innovation. Even more than in other engineering disciplines, standards matter, whether set by standards bodies or dominant players. Given the multi-year time frames of standards and the limited willingness of national funding bodies to support standardization work, this makes research impact harder, as does the increasing complexity of cellular networks and barriers to entry that shut out most researchers from contributing to large parts of commercial mobile networks.