We had a brief mention of http://gri.pe on The View a little while ago and we saw an impressive spike (short capture at http://www.youtube.com/watch?v=YsfuBeSLUCM). This particular show is on at least three times: once in the eastern time zone, once in mountain or central time and once on the pacific.
During the east-coast showing, our traffic spiked. The req/s graph had a 90˚ bend:
We saw a spike about half as big for the west-coast showing and the central showings were half as big again. The traffic from these spikes took a long time to decay: something on the order of days. Our base level of traffic has stayed much higher that it was before.
FWIW, our app is built on Appengine and it scaled us flawlessly. Once the traffic started, AppEngine automatically spun up instances. We peaked at 22 instances, which were automatically started and expired by AppEngine to keep us running:
- We had the ___domain http://gri.pe during the initial showing. From our data, a significant number of people watch The View live, and most watch during its first showing in the US.
- The URL is short and it was written on-screen (we always make sure to tell people to spell it out so they don't go to gripe.com).
- People likely watch "The View" for some form of education. They likely watch "The Middle" for passive entertainment. I imagine they would be more likely to explore something seen in the former than the latter.
I'm not sure the experiences are that different. mmastrac says his spike was during the broadcast. But the author of the post didn't own the ___domain during any of the broadcasts.
He watched the show a half an hour after it aired in his time zone (Central Time Zone). Meaning by the time he even watched the show it had already aired in every time zone but Pacific and he was watching it concurrently with that audience. Assuming he looked into it, registered the site and got it up within an hour or so he MIGHT have had it up for Alaska and Hawaii but that's about it.
I registered the ___domain at about 7:50pm Central time and had content on the site by 8:05; the show had aired at 7:00 Central time, so I was still two hours ahead of the Pacific time zone airtime (10pm Central time).
During the east-coast showing, our traffic spiked. The req/s graph had a 90˚ bend:
http://yfrog.com/755dmmp
We saw a spike about half as big for the west-coast showing and the central showings were half as big again. The traffic from these spikes took a long time to decay: something on the order of days. Our base level of traffic has stayed much higher that it was before.
FWIW, our app is built on Appengine and it scaled us flawlessly. Once the traffic started, AppEngine automatically spun up instances. We peaked at 22 instances, which were automatically started and expired by AppEngine to keep us running:
https://lh4.googleusercontent.com/_CqFM7LbtCbY/TNrJSS4B5qI/A...
Another interesting factoid: we paid around US$0.55 to handle the traffic from that spike.