DoorDash Processed More Than 10 Million Sweepstakes Entries on AWS

DoorDash is a local commerce platform that operates in more than 30 countries. To capture the attention of viewers, DoorDash launched a sweepstakes for one grand prize containing dozens of products seen in football championship TV ads, prompting viewers to enter a promotional code from its commercial on a microsite. With more than 100 million viewers, scalability and reliability were essential.

DoorDash powers most of its infrastructure using AWS because of its scalability and variety of services, and Wildlife has been building solutions on AWS since 2014. So, the companies chose to use AWS to build a scalable microsite for the sweepstakes. “We’ve invested heavily in using AWS services for nearly all our projects,” says Pavel Zagoskin, Technical Director at Wildlife.

For DoorDash’s sweepstakes website, Wildlife designed a multi–AWS Region EDA—an architecture pattern that is built from small, decoupled services that operate according to near real-time changes in state. “We chose a serverless approach because we didn’t want to deal with any servers or hardware management,” says Zagoskin. With a serverless architecture, the website would be able to handle millions of concurrent requests with a pay-as-you-go model.

DoorDash received support from AWS Countdown Premium, a service for optimizing business-critical events, launches, and migrations on AWS, 11 days before finishing the website. “It was invaluable to have dedicated support engineers from AWS,” says Brian Alexander, Technical Program Manager at DoorDash. “The engineers would get back to us very quickly with resolutions or suggestions.” The DoorDash and Wildlife teams worked closely alongside AWS throughout the project. “We saw the power and potential of collaborating with the experts at AWS and going beyond our knowledge for new or unfamiliar challenges,” says Jake Friedman, Cofounder and President of Wildlife.

Wildlife based the design on AWS Lambda, which businesses can use to run code without operating servers or clusters. To manage its API, DoorDash uses Amazon API Gateway, which makes it simple to create, maintain, and secure APIs at nearly any scale, across any number of Regions. DoorDash replicated its API across three Regions in the United States to distribute the load more evenly, provide redundancy, and avoid downtime. The company also used the latency-based routing feature of Amazon Route 53, a reliable and cost-effective way to route end users to internet applications, so that users would automatically be routed according to their proximity to the appropriate Region.

Wildlife and DoorDash performed stress tests until the night before the game, which revealed the importance of pre-scaling certain services to verify that the microsite was ready to deliver a fast, reliable user experience. “By pre-scaling the frequently used components in our load testing on recommendations from AWS, we felt more confident about the launch,” says Steve Hoffman, Software Engineer at DoorDash.

DoorDash connected Amazon API Gateway to Amazon Simple Queue Service (Amazon SQS), a fully managed message queue for microservices, distributed systems, and serverless applications, to decouple submission processing from the rest of the architecture. The company stored all data for the microsite using Amazon Simple Storage Service (Amazon S3), which provides object storage built to retrieve virtually any amount of data from anywhere.

To process the entries, DoorDash ran an AWS Lambda function to automatically retrieve data from Amazon SQS and send it to Amazon Kinesis Data Firehose, which reliably loads near real-time streams into data lakes, warehouses, and analytics services. “With the EDA, we had a lot of dashboards so that we could watch as the data moved through the environment to look for hot spots, both when we were testing and when the site was live,” says Hoffman. Meanwhile, to keep the frontend simple, the company took advantage of the distribution and scalability of Amazon CloudFront, which securely delivers content with low latency and high transfer speeds.