How DNS Load Balancing Distributes Multiple IPs for a Single Domain

One Domain, Multiple IPs

In theory a domain name can map to several IP addresses, allowing different users to reach different servers when they request the same hostname. To achieve this, you add multiple A records in your DNS provider, each pointing the domain to a distinct IP.

For example, to map www.itmtx.cn to three IPs—1.1.1.1 (Beijing Telecom), 2.2.2.2 (Shanghai Mobile), and 3.3.3.3 (Shenzhen Unicom)—you would create three A records:

www.itmtx.cn IN A 114.100.20.201;

www.itmtx.cn IN A 114.100.20.202;

www.itmtx.cn IN A 114.100.20.203;

When a Beijing user queries itmtx.cn, the DNS server returns one of the configured IPs based on the pre‑defined resolution policy, and a Shanghai user may receive a different IP. Each DNS query can yield a different address, forming a server cluster that provides load‑balancing, improves response speed, and increases availability through failover.

Common DNS Resolution Strategies

Typical DNS policies include:

Round Robin : Returns IPs sequentially, distributing requests evenly.

Weighted : Assigns a weight to each IP; higher‑weight servers are selected more often.

Response Time : Measures server latency and returns the fastest IP.

Different DNS products implement these strategies differently; for instance, DNSPod combines weight and random selection.

Pros and Cons of DNS‑Based Load Balancing

Advantages :

Offloads balancing to DNS, eliminating the need for dedicated load‑balancer hardware or software.

Simple, flexible, and low‑cost implementation suitable for most TCP/IP applications.

No code changes required on the application side.

Many DNS services support geo‑location routing, directing users to the nearest server.

Disadvantages :

DNS caching at multiple levels (recursive, authoritative, local) can delay propagation of IP changes, causing users to reach outdated servers.

Frequent cache refreshes to keep records up‑to‑date may increase DNS traffic.

DNS cannot actively monitor server load or dynamically adjust weights, so it lacks real‑time performance awareness; large sites often pair DNS with a secondary load balancer (e.g., least‑connections) for finer control.