Route 53 Latency Routing

Route 53 latency-based routing is a DNS routing policy that directs users to whichever available AWS region provides the lowest network latency for their specific location, rather than sending everyone to a single fixed region or splitting traffic by a fixed percentage.

How It Works

When latency-based routing is configured, Route 53 maintains latency measurement data between AWS regions and various points across the internet, and when a DNS query comes in, it evaluates the querying resolver’s approximate location and returns the IP address of whichever configured region currently offers the lowest measured latency to that location — rather than the geographically nearest region, since geographic proximity and actual network latency don’t always correlate perfectly due to internet routing paths, peering relationships, and undersea cable topology.

Why It’s Different From Geolocation Routing

It’s worth distinguishing latency-based routing from geolocation routing, a similarly named but different Route 53 policy: geolocation routing directs users based on their actual geographic location (often for legal or content-licensing reasons, like ensuring EU users are routed to EU-based infrastructure), while latency-based routing is purely about minimizing measured network latency, which may or may not align with geographic proximity.

Where It’s Used

Latency-based routing is a common choice for globally distributed applications with infrastructure deployed across multiple AWS regions, where the goal is simply “give each user the fastest possible experience” rather than any policy-driven requirement about which specific region serves which users. It’s frequently combined with health checks, so that a region experiencing an outage is automatically excluded from the routing decision even if it would otherwise offer the lowest latency to a given user.

Latency Data Sources and Update Frequency

Route 53’s latency measurements are derived from ongoing network performance data AWS collects between its own regions and various points across the internet, rather than measuring the latency of any individual end user’s specific connection in real time — this means the routing decision reflects generally representative latency patterns for a given network location rather than a live, per-request measurement, which keeps the DNS lookup itself fast while still making a reasonably well-informed routing choice. Because these underlying measurements can shift over time as network conditions and peering relationships change, the region that latency-based routing favors for a given part of the world isn’t necessarily fixed permanently, and can shift if AWS’s underlying network topology or measured performance changes.