Google App Engine, the company's platform-as-a-service, went down on Friday for many customers.
The cloud service began having problems on Friday afternoon. As of 16:44 BST, the service was reporting anomalies (pictured) for the performance of its three supported languages - Python, Java and Go - and was also showing problems for Memcache.
"We saw it a little bit more than an hour ago, when we saw that the ability rating was dropping pretty fast," Stephane Enten, vice president of services at European cloud-monitoring company Cedexis, told me at 16:50pm in London. "Here we saw that the ability was dropping to zero globally - it's pretty severe. I think it's very rare for these kind of services to go completely down. As of a few minutes ago, it started coming back."
Many companies tweeted that their own services were down as a consequence of Google App Engine's failure, highlighting the cat's cradle of interdependencies that modern cloud computing entails.
Services from companies such as A/B testing provider Optimizely, messaging and collaboration specialist Promevo and games company Humble were all affected as a consequence.
It is not unusual for major clouds to go down, and Google App Engine's failure can be read as another clear sign that, as Google says, "at scale, everything breaks".
As of 17:12pm, the service's status page was unavailable.