A misconfiguration on one of the gateway nodes caused requests to queue under load. Identified and corrected the configuration. Response times returned to normal immediately.

Some API requests experienced elevated response times and intermittent timeouts.