# How Nordstrom Migrated to Service Mesh with Kong Gateways

This talk explains how Nordstrom used its existing Kong gateway architecture and engineering standards to migrate large-scale Kubernetes workloads to service mesh without forcing risky application rewrites. Dakota Pollard says hundreds of teams and thousands of APIs first had to be brought onto a common traffic path through central and cluster gateways, with TLS removed from application containers and handled by sidecars, creating a "mesh light" foundation that later let those sidecars be swapped for a single mesh proxy. Nordstrom then "meshed the on-ramp" by enabling mesh first at the cluster gateway, so applications could cut over through a central gateway configuration change and roll back almost instantly if issues appeared. The approach was proven on Tax API, a latency-sensitive, multi-region checkout dependency, where an initial 503 problem was quickly reversed and fixed with Kong. Pard also highlights a remaining challenge around progressive canary rollouts, which Nordstrom manages with Gateway API and expects to extend into mesh through GAMMA support, and argues that platform ownership, standards, and fast rollback are what made production adoption possible.