Super Bowl Sunday is the time of year where most of America stocks up on light beer, snacks of questionable nutrient content, and heads over to their friend’s house with the brand new 72 inch 4K TV to watch two teams pit each other for the Vince Lombardi Trophy. Life360 users are no exception to this.
Life360 is a family safety app that focuses on driving safety and location awareness of your family members. Since the app is primarily in the background, Life360’s infrastructure engineering team has the unique prospect of having an “always on” traffic stream, and we oftentimes see variances in the traffic stream that correlate with national or worldwide events, e.g. New Year’s Eve, Christmas Day (or even the travel days prior and after Christmas Day), and the Super Bowl. Weather events are also quite visible.
Friday & Saturday Patterns
The screenshot above of our Grafana charts shows a typical traffic pattern for the Friday and the Saturday prior the Super Bowl. Friday has two apparent peaks: the morning commute and the evening commute, so we receive locations at a higher frequency than our normal off-peak times. Saturday is absent of these peaks as the traditional 9–5 workweek does not land on a Saturday.
Super Bowl Sunday
I’m showing the previous day for comparison — a typical Sunday is slightly lower than Saturday with the same peak profile, but here, you can see a very large gap and a very steep trending downward curve. You can also observe the slight uplift right before kickoff, which can be attributed to people rushing over to make the kickoff at 3:31PM PST. The game also ended at 7:17PM PST, but as you can see, the spike of traffic started well before the game was over, presumably because our Life360 users are forward-thinking and wanted to try to beat the traffic. Our autoscalers had to do a bit of extra work to accommodate a nearly 30% short term transient spike, but we weathered the event without affecting our API response times, thanks to the tireless work of the infrastructure team.
About the Technology
We primarily use Amazon Web Services to host our infrastructure, and we stream Cloudwatch data from our ELB’s to our Prometheus instances, where we can visualize the data in Grafana. Currently our ELB’s process over 2 billion requests per day.
We have many positions open for all of our platforms — backend, iOS, Android and infrastructure engineering. See our open positions!