Thanksgivingblackfridayads.com (TBFA) specializes in publishing leaked Black Friday Ads weeks before they’re printed in newspapers. Starting every year at the end of October, traffic spikes as people flood the site hoping to get an early peak of what deals they have to look forward to. For the rest of the year, the site sits basically idle.
Over the past three years, TBFA has consistently seen between 500%-1000% year-over-year growth. With 921,740 pageviews in November 2010, they were gearing up for around 5,000,000 pageviews in the month of November 2011. In the last week of October this year, TBFA began experiencing degraded site performance due to increased traffic. This was before their November rush had even started. They immediately began seeking possible solutions that would handle the impending surge.
According to Scott Offord, the site owner, “We were exploring the option of using Amazon EC2, but decided we wanted another layer above that. We wanted something more managed than just going straight to the raw source.” Pagoda Box provided the simple server management, granular pricing and control they were looking for.
TBFA runs on a WordPress install that utilizes optimization tools like WP Super Cache. Combined with the distributed scaling model of Pagoda Box, TBFA handily served 5.9 million pages to 413,458 visitors between November 1st and November 30th, 2011. At the height of their traffic spike, they served approximately 600 concurrent users up to 900 pages per minute.
Just as TBFA moved over to Pagoda Box, traffic began to increase. On a single, free 200 MB RAM web instance and a 200 MB RAM MySQL DB, TBFA served approximately 160,000 web requests in a 24-hour period, as covered in The Battle of the Single Instance Apps.
Over the month of November, TBFA had up to 500,000 per day. With the visibility provided by built-in app analytics and the ability to seamlessly scale when needed, TBFA’s customers never saw an increase in average page load times, even when the site was under heavy load.
For the last 3 years, TBFA has seen between 500% and 1000% growth each year. After moving to Pagoda Box and by utilizing the concurrency scaling model, this kind of explosive growth can be managed without major alterations of their application codebase.
Pagoda Box’s granular scaling model allowed TBFA to scale to meet their demand without overbuilding their infrastructure. Instead of purchasing and/or beefing up servers and crossing their fingers, hoping it was enough, they were able to scale to match the ebbs and flows of their traffic, drastically reducing the cost of their infrastructure overhead.