How Internet Exchange Points Work
Internet Exchange Points (IXPs) are physical locations where networks connect to exchange traffic directly. Instead of routing through multiple transit providers, networks peer at IXPs to reduce latency, cut costs, and keep local traffic local.
At a typical IXP, hundreds of networks—ISPs, content providers, enterprises, and cloud platforms—connect their routers to shared switching infrastructure. When Network A wants to send data to Network B, they can exchange it directly across the IXP fabric rather than paying a third party to carry it.
The largest IXPs handle tens of terabits per second of peak traffic. DE-CIX Frankfurt and IX.br São Paulo regularly exceed 17 Tbps, representing a significant portion of global Internet traffic.
Submarine Cable Infrastructure
Over 95% of intercontinental Internet traffic travels through submarine fiber optic cables lying on the ocean floor. These cables, typically about the diameter of a garden hose, carry the vast majority of global data.
Modern submarine cables use wavelength-division multiplexing to transmit hundreds of terabits per second through multiple fiber pairs. The newest transoceanic cables can carry 400+ Tbps—enough for billions of simultaneous video calls.
Cable landing stations, where submarine cables come ashore, are critical infrastructure points. Major hubs like Singapore, Marseille, and Virginia Beach connect dozens of cables, making them strategically vital to global connectivity.
A single cable cut (from ship anchors, earthquakes, or equipment failure) can reroute significant traffic, though redundancy has improved. The 2008 Mediterranean cable cuts affected Internet access for millions, accelerating investment in diverse routing.
Peering Economics and Tier 1 Networks
Settlement-free peering: When two networks agree their traffic is roughly equal, they exchange it freely without payment. This "handshake" arrangement drives IXP growth and reduces costs for both parties.
Paid peering: When traffic ratios are unbalanced (typically content providers sending more than they receive), the heavier sender may pay the receiver. Netflix and other streaming services often negotiate paid peering arrangements.
Transit: Smaller networks pay larger carriers for connectivity to destinations they can't reach through peering. Transit providers guarantee reachability to the entire Internet.
Tier 1 networks: A handful of networks (Lumen, NTT, Cogent, Telia, GTT, etc.) peer freely with each other and need no transit—they can reach anywhere through their mutual agreements. Everyone else ultimately pays someone for connectivity.
Understanding Global Latency
Speed of light limits: Light travels through fiber at roughly 200,000 km/s (two-thirds the vacuum speed). Crossing the Atlantic (6,000 km) takes at least 30ms one-way—physics imposes hard latency floors regardless of network quality.
Why location matters: Content delivery networks (CDNs) cache data at the edge specifically to avoid these latency penalties. A video buffered from a local CDN node loads instantly; fetching it from another continent adds noticeable delay.
Regional Internet growth: Investment in local IXPs and content caching has dramatically improved Internet quality in emerging markets. African latency to global content dropped significantly as major platforms deployed local points of presence.
Gaming and real-time applications: For interactive applications, even 50ms round-trip time affects user experience. This drives demand for distributed infrastructure and explains why cloud gaming services need data centers in every major region.