At 2:44pm Pacific, we received a report of slow cart loads for one of our users. We were able to trace the cause to a load balancer (AWS ELB) that was having much higher latency than it should have.
From our testing, this failing ELB was causing occasional requests (not all, probably 5%) to take over a minute for the initial connection.
As of 3:35pm, we replaced the ELB, and are continuing to monitor the situation.
4:07pm Update: We have improved our monitoring to better detect anomalies in our load balancers in the future.