Routing Policies

~12 min read

Simple, Weighted, Latency-based, Failover, Geolocation, and Multi-value — six ways to decide which endpoint a request goes to.

Simple routing sends all traffic to a single resource — no rules, the default for a single-endpoint domain. Weighted routing splits traffic across multiple resources by a percentage you assign, commonly used for canary deployments (send 5% of traffic to a new version) or gradual migrations.

Latency-based routing sends users to whichever configured Region gives them the lowest network latency, based on AWS's continuously measured latency data between end-user locations and AWS Regions — not simple geographic proximity, which doesn't always correlate with actual network latency. Geolocation routing instead routes based on the user's actual geographic location (country/continent/state), useful for content licensing restrictions or legally-required data residency routing rather than performance optimization.

Failover routing implements active-passive DR: a primary record serves all traffic while its health check passes; if it fails, Route 53 automatically switches all traffic to the secondary record. Multi-value routing returns multiple healthy IP addresses in response to a single query (up to 8), giving basic client-side load distribution and health-check-based exclusion of unhealthy endpoints, without the sophistication of a real load balancer.

💬 Deep Dive with AI

Key points

  • Simple: one endpoint, no rules — the default
  • Weighted: percentage split, great for canary releases and gradual migrations
  • Latency-based: routes by measured network latency, not simple geographic distance
  • Geolocation: routes by user's actual location, for compliance/licensing, not performance
  • Failover: active-passive, health-check-driven — the DR pattern
  • Multi-value: returns several healthy IPs per query for basic distribution and health filtering