Amazon Web Services's main cloud computing service, EC2, is having problems, which means businesses and developers could have trouble accessing their rented compute resources.
The network connectivity issues with EC2 began at 10:08am (2:08am PDT) on Monday and were still in place at lunchtime.
"We are continuing to work to resolve the networking connectivity issues resulting in elevated packet loss for some EC2 instances in the US-EAST-1 region," Amazon said on its AWS global status page.
US-EAST-1 is Amazon's main datacentre hub. It provides services to a large number of websites, including Reddit, which at the time of writing was intermittently inaccessible.
"Well at least downtime like this gives you chance to work out where your system fails elegantly, and where it fails like a beast," Twitter user Jamie Jin wrote.
As US-EAST-1 is Amazon's main datacentre region it is the one most prone to problems. The last major failure out of this region occurred in July when a series of hidden bugs meant a standard outage spiralled into a cascading fail, hitting scores of businesses.