Tracing the Path to YouTube: A Quantification of Path Lengths and Latencies Toward Content Caches

Abstract

Quantifying the benefits of content cache deployments, both in terms of latency and path lengths, and identifying the disparity of achievable benefits over IPv6 and over IPv4, is essential to identify bottlenecks in content delivery. We approached this matter by collecting and analyzing traceroute measurements toward YouTube from ~100 vantage points located in 74 different origin autonomous systems (ASs). Using a longitudinal dataset (05/2016–05/2018), we show that most of the Google Global Cache (GGC) nodes were reachable within ~6 IP hops and within ~20 ms from users streaming the videos. Further, we observed that in cases where GGC nodes were dual-stacked, the path lengths and latencies were comparable over both address families. However, as generally believed, shorter path lengths did not always correlate with lower latency: when the video was cached by a GGC node over IPv6 only, paths were almost always shorter over IPv6. Yet, latencies were still lower over IPv4, indicating room for improvement over IPv6. GGCs reduced IP path lengths and latencies by up to a third over IPv4 and by up to a half over IPv6, stressing the importance of content cache deployments in ISP networks. To encourage reproducibility of this work, we make the entire dataset available to the community.