Filed under: Performance Monitoring
So we all know that the Baltimore Ravens are the 2013 Blackout…err…Super Bowl Champions. And debate will rage on about the “winners” of those Super Bowl commercials. Running away with the social media title this year was, without a doubt, Twitter. With the power outage that almost brought the Lombardi trophy to San Francisco (as a Steeler fan, I guess I’m happy to see it in Baltimore?), Twitter was able to capture fan reaction and to bring a diminishing viewer base back in front of their TVs and the edge of their seats to watch one of the greatest finishes since Ben Roethlisberger to Santonio Holmes in Super Bowl XLIII. (On a side note, CBS should in some way ante up some cash to the social media giant for making this the 3rd most-watched Super Bowl in history. But I digress.)
How was Twitter able to handle all that traffic? The halftime show had generated some Beyonce buzz, on average 268,000 tweets per minute, but clearly, with an early 3rd quarter kick-off return of 108 yards adding to an already insurmountable lead for Baltimore, the game didn’t appear to be much in doubt and viewers were starting to tune out. Enter the great black out of 2013.
We (namely this AppNeta Systems Engineer) had the foresight to establish some path monitoring to the Twitter-verse to look at real-time traffic utilization during the Super Bowl. I assumed tweets would be fast and furious, as I had expected a close game. Only when setting this up, I didn’t realize how pivotal this would end up being.
The charts below indicate specific data points during the black out. From the initial shut down of power (right around 9pm) to the point when referee Jerome Boger said over his mic “Let’s go!” 34 minutes later, tweeters were averaging 231,500 tweets a minute during the outage. We were able to determine that all that tweeting didn’t have much of an impact on Twitter’s overall network performance. Basically, if you had a thought you wanted to share, it was going to reach the masses.
The above chart shows a data point early in the black out. An increase in traffic along the network path (all the excitement of a half dark stadium) rose to a peak of 54% of the available capacity, but didn’t really taxing the network.
This screen grab shows a data point just after halfway through the blackout. There was pretty steady utilization for almost a 10-minute period, but again, it wasn’t really taxing the network. Fans at the game were tweeting about the small factions they had formed for survival as they fought off roving bands trying to take their seats. #revolution
This third screen grab shows a data point late in the black out. People were getting anxious – was the game ever going to come back? Utilization was higher than what it had been over the course of the black out. Those in the nosebleed sections at the Superdome were wondering when the time would come to exit stage right to avoid the rush.
What have we learned about the Twitter-verse during the blackout?
- They clearly were ready for the onslaught of the Super Bowl, regardless of the outcome.
- They pay their bills. Well played Twitter, well played.