View In:   Map Viewer

The Measurement Lab, or M-Lab, project hosts the largest open database of internet speed and connectivity data in the world. It was established in 2009 by New America’s Open Technology Institute, the PlanetLab Consortium, Google, and a group of academic institutions to enable better internet connectivity research. In 2019, it was acquired by the nonprofit organization Code for Science and Society. PlanetLab shut down in May 2020, but Google still provides server hosting and other services. M-Lab remains an open-source project, and its code is available to anyone to use.

M-Lab collects internet performance data using several user-driven tools. This layer draws on data collected from the Network Diagnostic Tool (NDT), accessible here, which records upload and download speeds, as well as latency and other metrics, by uploading a large test file to a server. Collected data is hosted in a cloud storage platform, where it can be queried and gathered by researchers. M-Lab data differs from FCC BDC data in that it does not rely on self-reporting by Internet Service Providers. Because its testing algorithm focuses on single-stream download performance, M-Lab data generally better describes the actual nature of a user’s internet experience in comparison with BDC and Ookla. This method can take into account factors like the type of machine, other programs running, and other users taking bandwidth on a shared network. However, it often does not reflect the speeds that would be possible were it not for network usage and hardware/software limitations, leading to criticism from Ookla that it understates available speeds in many areas.

The M-Lab 2023 layer contains M-Lab download and upload speed data from NDT tests over the period 1/1/2023 - 12/31/2023, aggregated to two sublayers: US Counties and US States. The popup windows feature a report that provides more information on M-Lab results in that area.

Annual statistics are determined by selecting random results for each IP address per day by geography, then calculating daily min/max/median results by geography from the random sample of results. Data for upload and download are then combined and min/max/median statistics for the year are generated by geography from the daily summaries.