This issue was caused by a load spike on a primary datastore disk which began at 09:41 and was causing requests to complete slower than normal.

At 09:47 the cause was identified and blocked from causing further load. This immediately improved the speed of responses to our API.

At 09:49 additional capacity was brought online as a precaution, although normal levels of service had already been restored.

We apologise for the brief slowdown.

Normal behaviour restored, but continuing to investigate

Investigating