AWS hit by overheating outage in northern Virginia, disrupting Coinbase

"Amazon Web Services said on Thursday that one of its data centres in northern Virginia was running hot enough to disrupt customer workloads, and that engineers were still bringing the site fully back online when most users had gone to bed for the night."

"The trigger was prosaic: increased temperatures inside a single data centre, attributed to a cooling-system shortfall, forced AWS to throttle and then partly reroute traffic away from the affected Availability Zone."

"By the company's account, additional cooling capacity began coming online a couple of hours after the first impact reports, and "early signs of recovery" appeared shortly after. A later update was less reassuring: bringing in enough extra cooling to safely restart the remaining systems was taking longer than expected, and AWS was unwilling to put a clock on full restoration."

"Coinbase confirmed that its trading platform problems were caused by the AWS event. After several hours of degraded markets, the exchange said all markets had been re-enabled, and trading was back to normal. C ME Group, the world's largest derivatives marketplace, also reported issues with its CME Direct platform during the same window, although it described the cause only as "essential maintenance" and did not say whether the AWS event was a factor."

A northern Virginia AWS data centre experienced increased internal temperatures due to a cooling-system shortfall. AWS throttled and partly rerouted traffic away from the affected Availability Zone to reduce impact on customer workloads. Additional cooling capacity came online a couple of hours after initial reports, and early signs of recovery appeared shortly afterward. Restoring enough cooling to safely restart remaining systems took longer than expected, and AWS did not provide a timeline for full restoration. Coinbase reported that its trading platform problems were caused by the AWS event, with markets degraded for several hours before being re-enabled and returning to normal. C ME Group reported issues with its CME Direct platform during the same window, attributing them to essential maintenance without confirming a link to the AWS event.

Read at TNW | Amazon