DNS Resolution

How it works

DNS resolves human-readable domain names into IP addresses through a hierarchical lookup chain — a recursive resolver, root servers, TLD servers, and finally the domain's authoritative name server. Once resolved, the answer is cached according to its TTL, so almost every real-world lookup skips the full chain.

Mental models

• Resolution walks a hierarchy: root → TLD (.com) → authoritative name server, each pointing to the next.
• A resolver caches the answer for the record's TTL — most lookups never leave the cache.
• A records map a name to an IP; CNAME records alias one name to another; NS records delegate a zone to its name servers.
• Weighted round robin, latency-based, and geolocation-based routing let a single domain answer differently per client.
• DNS runs on eventual consistency at internet scale — propagation delay after a change is normal, not a bug.

Common pitfalls

• Every uncached lookup adds a full round trip before the 'real' request can even start — TTL tuning is a real latency lever.
• Lowering TTL for faster failover means more traffic hits your authoritative servers and more exposure to their outages.
• DNS is a centralized dependency — an outage or DDoS against your resolver takes down every service behind that domain, even if the services themselves are healthy.