Today the API experienced intermittent outages from 8:50 ET to 9:00 ET, and from 11:00 ET to 11:30 ET. This outage was caused by a large denial-of-service attack on our service provider, Linode, which caused a shortage of resources (specifically, CPU time and network bandwidth) to our host network. This manifested as greatly slowed access to the Forecast website and API, or, intermittently, no access at all.
During the second downtime window, we took several services offline in order to perform some network reconfiguration, which greatly relieved resource tension on our hosts and prevented any further outages. Furthermore, Linode is performing additional analysis and working to strengthen their network against these kinds of attacks, so further downtime of this kind should be unlikely.