How DNS Servers Route Web Traffic to the Least Busy Server

A method for balancing web traffic by having servers report their current workload to a DNS server, which then directs new users to the fastest available machine.

What it covers

This patent describes a system where multiple web servers share a single domain name (like www.example.com). Each server monitors its own performance by measuring how long it takes to respond to requests and how many clients are currently waiting. The servers send these performance metrics to a central DNS server. When a new user tries to visit the site, the DNS server uses this data to calculate which server is currently the least busy and sends the user to that specific machine's network address.

What it doesn't cover

• —Does not cover load balancing performed by hardware appliances like dedicated load balancers sitting in front of servers.
• —Does not cover client-side selection where the user's browser decides which server to connect to.
• —Does not cover simple round-robin DNS where servers are chosen in a fixed sequence regardless of their actual performance.

The clever bit

The system shifts the intelligence of load balancing into the DNS lookup process itself, using real-time performance feedback from the servers to make routing decisions before the user even connects.

Why it matters

This technology was essential for the early scaling of the internet. Before this, popular websites would crash if one server became overwhelmed. By using DNS to intelligently distribute traffic, companies could keep services online even when user demand spiked, laying the groundwork for modern cloud infrastructure.

Real-world examples

• 1.Early web server clusters
• 2.Global Server Load Balancing (GSLB) systems
• 3.Content Delivery Network (CDN) routing logic