DNS Resolution Per-Request Overhead

warning

latencyUpdated Sep 7, 2025

Kong resolves upstream hostnames for every request by default, adding 20-100ms latency per request. This DNS lookup overhead becomes a severe bottleneck at high request rates.

Sources

Kong Gateway Performance Optimization - When Your API Gateway Becomes the Bottlenecktoolstac.com

Technologies:

Kong GatewaySymptoms of this issue are visible in Kong Gateway metrics and logs

kong.upstream.connection_time.p99

kong.dns.query.duration.p99

kong.latency.p99

CoreDNSCoreDNS metrics correlate with this issue and help confirm diagnosis

How to detect:

Enable Kong debug logging and check request timing breakdown for DNS resolution time. If upstream connection time is high (20-100ms) despite local backends, or if Kong latency is disproportionately higher than upstream latency, DNS resolution is likely the cause. Monitor for DNS query rate matching request rate.

Recommended action:

Use IP addresses instead of hostnames in upstream configurations. If hostnames are required, increase DNS TTL to reduce lookup frequency. Consider deploying local DNS caching (dnsmasq or CoreDNS) to minimize external DNS query latency. Set dns_stale_ttl to allow Kong to use stale DNS entries during resolution failures.