You Get What You Measure: Internet Performance Measurement as a Policy Tool

Abstract

The research literature on Internet performance measurement is quite rich. Surveys of measurement tools such as “A Study of Traffic Management Detection Methods & Tools” and “A Survey on Internet Performance Measurement Platforms and Related Standardization Efforts” describe a multitude of tools such as NetPolice, NANO, DiffProbe, Glasnost, ShaperProbe, Chkdiff, Sam Knows, BISmark, Dasu, Netradar, Portolan, RIPE Atlas, and perfSONAR.

In addition to tools developed for academic research and policy enforcement, Internet users rely on Speedtest and OpenSignal for troubleshooting. Finally, proprietary systems such as those developed by Akamai, Sandvine, and Cisco are used to compile “State of the Internet” analyses aggregating several views of the Internet.

While current tools are quite useful for measuring the performance of broadband networks, they’re much less useful for examining how well the Internet operates as a whole. The Internet is an “end-to-end network of networks” in which performance depends on a series of cooperating networks as well as network-attached devices and services.

From the user perspective, the web appears to be slowing down. While this trend has become received wisdom, traditional measures of broadband performance continue to show improvement: Akamai’s measurements of “average peak connection speed” show US average speed increased an average of 29 percent per year between 2010 and 2017.

The emphasis on one facet of Internet performance, such as last mile broadband networks, tends to minimize other factors that may be more important to the user, such as browsers, web page design, and webserver performance. In addition, a reliance on active measurement tools creates opportunities for gaming the system that are not possible in passive systems that merely observe application and network events in real time. Passive systems have privacy issues, however.

This paper explores opportunities for developing additional performance tools more responsive to the broader social goal of better end-to-end Internet performance across the broad span of applications. It finds that a system for capturing passive measurements and sharing them between ISPs, web developers, and other responsible parties may be useful for accelerating the web experience.

Keywords: Internet, performance measurement, policy enforcement, end-to-end, network of networks, broadband

JEL Classification: L88, C89, C93