Understanding Request Routing in Global Networks

Every HTTP request is a small routing problem. The browser knows a hostname; it needs an IP address and a physical path through the internet to reach a server. Several layers of routing infrastructure solve this, usually in under 100ms.

DNS: translating names to IPs

Before a TCP connection opens, the browser resolves the hostname to an IP via DNS. For global services, DNS itself does routing work: GeoDNS returns different IPs based on the requester's location, directing users to the nearest PoP.

BGP: routing at the network level

Once the IP is known, packets travel through the internet via BGP (Border Gateway Protocol). Each AS (Autonomous System, like an ISP or cloud provider) announces which IP ranges it can reach. Routers along the path pick the best next hop based on BGP attributes.

Anycast: one IP, many destinations

With anycast, the same IP is advertised from multiple locations. BGP naturally routes each packet to the closest advertising location. This is how CDNs and DDoS mitigation networks achieve low-latency global reach with a single IP address.