"Rush hour" slowdowns which affected the service daily since Tuesday, did not occur today. During the previous hours, we've made changes to our Load balancer's scaling configuration with the help of AWS support, tested and confirmed that the changes work. Now that the critical load hours have passed, we conclude that the solution worked.
Posted Jan 15, 2016 - 14:28 EST
Update
We have made progress in resolving the recurring ELB slowdowns we had in the last few morning "rush hours" (between 16:00-19:00 UTC). We continue to closely monitor the system around the clock to prevent re-occurrence, and are working closely with the technical team at AWS to match the capacity they provide us and our system congurations to the traffic at these times.
Posted Jan 15, 2016 - 08:39 EST
Monitoring
Between 18:05 and 18:35 UTC on Jan 14, customers experienced slowdowns in image delivery, similar to yesterday's slowness, but less severe and for a shorted time frame. The steps we took yesterday to mitigate errors have prevented uploads from being affected, however average image delivery time did increase from 0.5 to 1.5 seconds. As of 18:35, we are back to normal response rates, and continuing to monitor the service and work with AWS support on a long-term solution to the ELB issues we are experiencing.