Today (12th of June, 2015) many ISPs and end users may have experienced increased packet loss and general performance problems. The issue started around 10:00 UTC and lasted for around one hour (followed by another hour period during which BGP transitioned back to stable state). It seems that the problem was caused by one of the ISPs advertising a very large number of BGP routes to a Tier-1 network. Because of that plenty bogus BGP updates were triggered across pretty much all routers in the world:
Since the ISP advertised far more routes than they were able to handle - packets going to this ISP were essentially hitting a black hole. The rather small ISP simply couldn't process the surge of traffic.
The routes advertised by ISP had a rather small cost associated with them which likely caused a disruption in traffic globally. So if you're wondering where did this spike in your Keynote/Gomez test come from - well, this is probably the reason.