interesting though, could be a catching problem or latency at the end user or lag at the isp. i would assume those servers are in redundant clusters around the us and perhaps globe. a loss in a primary isp to the a core set could cause this issue. it would interesting to know where the primary data center is for this