Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Investigating Web Content Delivery Performance over Starlink

Rohan Bose, Jinwei Zhao, Tanya Shreedhar, Jianping Pan, Nitinder Mohan

TL;DR

This paper addresses how Starlink's LEO network affects web content delivery by decomposing end-to-end latency across PoP, DNS, and CDN layers. It leverages a two-year, multi-layer measurement campaign combining Cloudflare AIM, M-Lab, RIPE Atlas probes, and controlled Starlink vantage points to quantify satellite and terrestrial contributions to latency. The study identifies three performance regimes driven by infrastructure density, demonstrates that PoP proximity often dominates, and reveals that DNS/CDN misalignment and cache misses can cause large penalties, particularly in Africa. It shows that strategically relocating PoPs yields substantial gains where CDN maturity is low, and it argues for satellite-aware request mapping and potential in-orbit caching, with open data and code to support reproducibility and future work.

Abstract

Low Earth Orbit (LEO) satellite ISPs promise universal Internet connectivity, yet their interaction with content delivery remains poorly understood. We present the first comprehensive measurement study decomposing Starlink's web content delivery performance decomposed across Point of Presence (PoP), DNS, and CDN layers. Through two years of measurements combining 225K Cloudflare AIM tests, M-Lab data, and active probing from 99 RIPE Atlas and controlled Starlink probes, we collect 6.1M traceroutes and 10.8M DNS queries to quantify how satellite architecture disrupts terrestrial CDN assumptions. We identify three distinct performance regimes based on infrastructure density. Regions with local content-rich PoPs achieve near-terrestrial latencies with the satellite segment dominating 80-90% of RTT. Infrastructure-sparse regions suffer cascading penalties: remote PoPs force distant resolver selection, which triggers CDN mis-localization, pushing latencies beyond 200 ms. Dense-infrastructure regions show minimal sensitivity to PoP changes. Leveraging Starlink's infrastructure expansion in early 2025 as a natural experiment, we demonstrate that relocating PoPs closer to user location reduces median page-fetch times by 60%. Our findings reveal that infrastructure proximity, not satellite coverage, influences web performance, requiring fundamental changes to CDN mapping and DNS resolution for satellite ISPs.

Investigating Web Content Delivery Performance over Starlink

TL;DR

This paper addresses how Starlink's LEO network affects web content delivery by decomposing end-to-end latency across PoP, DNS, and CDN layers. It leverages a two-year, multi-layer measurement campaign combining Cloudflare AIM, M-Lab, RIPE Atlas probes, and controlled Starlink vantage points to quantify satellite and terrestrial contributions to latency. The study identifies three performance regimes driven by infrastructure density, demonstrates that PoP proximity often dominates, and reveals that DNS/CDN misalignment and cache misses can cause large penalties, particularly in Africa. It shows that strategically relocating PoPs yields substantial gains where CDN maturity is low, and it argues for satellite-aware request mapping and potential in-orbit caching, with open data and code to support reproducibility and future work.

Abstract

Low Earth Orbit (LEO) satellite ISPs promise universal Internet connectivity, yet their interaction with content delivery remains poorly understood. We present the first comprehensive measurement study decomposing Starlink's web content delivery performance decomposed across Point of Presence (PoP), DNS, and CDN layers. Through two years of measurements combining 225K Cloudflare AIM tests, M-Lab data, and active probing from 99 RIPE Atlas and controlled Starlink probes, we collect 6.1M traceroutes and 10.8M DNS queries to quantify how satellite architecture disrupts terrestrial CDN assumptions. We identify three distinct performance regimes based on infrastructure density. Regions with local content-rich PoPs achieve near-terrestrial latencies with the satellite segment dominating 80-90% of RTT. Infrastructure-sparse regions suffer cascading penalties: remote PoPs force distant resolver selection, which triggers CDN mis-localization, pushing latencies beyond 200 ms. Dense-infrastructure regions show minimal sensitivity to PoP changes. Leveraging Starlink's infrastructure expansion in early 2025 as a natural experiment, we demonstrate that relocating PoPs closer to user location reduces median page-fetch times by 60%. Our findings reveal that infrastructure proximity, not satellite coverage, influences web performance, requiring fundamental changes to CDN mapping and DNS resolution for satellite ISPs.
Paper Structure (19 sections, 14 figures, 2 tables)

This paper contains 19 sections, 14 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Measurement Methodology
  3. Passive Measurements
  4. Active Measurements
  5. Global Content Delivery Overview
  6. CDN Performance
  7. DNS Performance
  8. Case Studies
  9. African Infrastructure Expansion
  10. Global PoP Reassignment
  11. Related Work
  12. Conclusion
  13. Ethics Statement
  14. CDN-Hosted Website Target List
  15. Additional Analysis
  16. ...and 4 more sections

Figures (14)

  • Figure 1: Median RTT difference (Starlink $-$ Terrestrial) between Starlink subscribers and Terrestrial subscribers to the best Cloudflare CDN server reported by Cloudflare AIM.
  • Figure 2: TN subscribers reach local CDN edges hosting locally relevant content, while LSNs must first traverse to distant PoPs before accessing CDN infrastructure. CDN mapping techniques select caches near the PoP resulting in cache misses and increased latency to fetch content from origin.
  • Figure 3: CDN latency breakdown over Starlink from selected countries: median minRTT to CDN caches (in orange) and DNS resolvers (in blue). Hatched sections isolate the PoP segment, with percentages marking end-to-end RTT share.
  • Figure 4: Access latencies from Starlink subscribers to CDN servers of Cloudflare, Akamai and CloudFront.
  • Figure 5: Provider-specific CDN mapping across Starlink. Each bar shows the distribution of CDN server locations selected by the provider's mapping (labels inside bars) across Cloudflare (left), Akamai (center) and Cloudfront (right).
  • ...and 9 more figures