We experienced an unexpected burst of event traffic this morning from approximately 9:00-11:30 am PDT. Normally we operate with enough capacity to handle these types of spikes, but in this case we started to fall behind. We have identified two different factors that seem to have contributed to the apparent reduction in peak throughput. One is related to a de-duping mechanism (a process which ensures we don't save the same event twice), and the other is related to proper even distribution of incoming events in our queue layer. We are working to investigate both of these and hopefully ensure a higher ceiling for future bursts. We are also in contact with the customers involved in the traffic burst to try and better coordinate on future usage.