Search for: All records

We investigate network peering location choices, focusing on whether networks opt for distant peering sites even when nearby options are available. We conduct a network-wide cloud-based traceroute campaign using virtual machine instances from four major cloud providers to identify peering locations and calculate the “peering stretch”: the extra distance networks travel beyond the nearest data center to their actual peering points. Our results reveal a median peering stretch of 300 kilometers, with some networks traveling as much as 6,700 kilometers. We explore the characteristics of networks that prefer distant peering points and the potential motivations behind these choices, providing insights into digital sovereignty and cybersecurity implications. 

We present a longitudinal study of intercontinental long-haul links (LHL) - links with latencies significantly higher than that of all other links in a traceroute path. Our study is motivated by the recognition of these LHLs as a network-layer manifestation of transoceanic undersea cables. We present a methodology and associated processing system for identifying long-haul links in traceroute measurements, and report on our findings from. We apply this system to a large corpus of traceroute data and report on multiple aspects of long haul connectivity including country-level prevalence, routers as international gateways, preferred long-haul destinations, and the evolution of these characteristics over a 7 year period. 

We present a longitudinal study of intercontinental long-haul links (LHLs) - links with latencies significantly higher than that of all other links in a traceroute path. Our study is motivated by the recognition of these LHLs as a network-layer manifestation of critical transoceanic undersea cables. We present a methodology and associated processing system for identifying long-haul links in traceroute measurements. We apply this system to a large corpus of traceroute data and report on multiple aspects of long haul connectivity including country-level prevalence, routers as international gateways, preferred long-haul destinations, and the evolution of these characteristics over a 7 year period. We identify 85,620 layer-3 links (out of 2.7M links in a large traceroute dataset) that satisfy our definition for intercontinental long haul with many of them terminating in a relatively small number of nodes. An analysis of connected components shows a clearly dominant component with a relative size that remains stable despite a significant growth of the long-haul infrastructure. 

The main premise of this work is that since large cloud providers can and do manipulate probe packets that traverse their privately owned and operated backbones, standard traceroute-based measurement techniques are no longer a reliable means for assessing network connectivity in large cloud provider infrastructures. In response to these developments, we present a new empirical approach for elucidating private connectivity in today's Internet. Our approach relies on using only "light-weight" ( i.e., simple, easily-interpretable, and readily available) measurements, but requires applying a "heavy-weight" or advanced mathematical analysis. In particular, we describe a new method for assessing the characteristics of network path connectivity that is based on concepts from Riemannian geometry ( i.e., Ricci curvature) and also relies on an array of carefully crafted visualizations ( e.g., a novel manifold view of a network's delay space). We demonstrate our method by utilizing latency measurements from RIPE Atlas anchors and virtual machines running in data centers of three large cloud providers to (i) study different aspects of connectivity in their private backbones and (ii) show how our manifold-based view enables us to expose and visualize critical aspects of this connectivity over different geographic scales.