Southwest Airlines Scales App Development Using AWS Device Farm & Amazon EC2 Mac Instances

Southwest flies 130 million passengers a year to 120 airports in 11 countries. Facilitating a great travel experience for such a high number of customers is challenging. That is why Southwest focuses on efficient operations and self-service options. Today, its fliers can use both the desktop and mobile applications to book, change, and cancel flight, car, and hotel reservations and upgrade boarding. Soon, they will be able to track their bags and receive more real-time travel updates. Customers are increasingly using self-service.

Expanding the functionality of Southwest’s applications means likewise scaling their release process. Previously, testing had to be manually performed on a physical device; the company used numerous smartphones, tablets, and Mac computers to test every application change. Faulty devices often resulted in testing delays caused by waiting for new ones. Moreover, the COVID-19 pandemic made swapping and procuring devices more difficult.

Southwest needed a solution that would facilitate automated, test-driven development and avoid the handoff of devices between testing teams. It chose AWS Device Farm, an application testing service that organizations use to improve their web and mobile apps by testing them across a range of desktop browsers and real mobile devices without having to provision and manage a testing infrastructure. AWS Device Farm maintains physical devices on behalf of Southwest, automatically confirming that each device is functional and ready for tests on demand. “AWS Device Farm is very user-friendly and easily accessible,” says Sundeep Bobba, cloud and DevOps tech lead on Southwest.com digital platforms.

Southwest began using AWS Device Farm in 2019 and uses it today for test automation across various platforms. From AWS Device Farm, Southwest can use public and private devices with ease. These devices are exclusively for Southwest’s use and provide connectivity to the company’s on-premises environment. AWS Device Farm also generates reports, including screenshot videos that testers can save as documentation.

Although the Southwest team is now using AWS Device Farm devices for manual testing, it is also writing application-performance-monitoring pipelines on GitLab to automate testing on the devices. Using GitLab and AWS Device Farm, teams can share test functions, centralize results, and derive analytics on how to improve their overall testing function as much as possible.

Southwest also paired AWS Device Farm with Appium, an open-source automation tool for running functional tests on devices. That accelerated test-run speed by an average of 7.1 times for new features: 2.3 times for iOS, 2.5 times for Android, 6 times for iPadOS, and 15 times for browsers. The company uses AWS Device Farm to simply scale its automation test workloads of Appium tests to dozens of device types and operating systems and versions of those systems.

Southwest did away with its entire fleet of Mac computers by 2021 and now relies on Amazon Elastic Compute Cloud (Amazon EC2) Mac Instances, which organizations use to run on-demand macOS workloads in the cloud. “There was huge applause when we achieved that, because we can now run and manage builds on demand,” says Bobba. Now, the Southwest team can automatically provision Amazon EC2 Mac Instances in minutes.

After migrating the Apple iOS CI/CD pipeline to Amazon EC2 Mac Instances, build speed increased by four times. Using the more scalable, flexible infrastructure on AWS, Southwest supported the same development lifecycle for its mobile and desktop apps. That gives customers the same experience on a browser as on the mobile app.

On AWS, Southwest can accelerate the delivery of new features to customers. “Using Amazon EC2 Mac Instances and AWS Device Farm, we find and fix issues earlier in the pipeline,” says Mona Pinjani, senior manager of technology at Southwest. “That saves cost because fixing errors later in production is expensive.”

Southwest also improved security by using Amazon EC2 Mac Instances and AWS Secrets Manager, which organizations can use to manage, retrieve, and rotate database credentials, API keys, and other secrets throughout their lifecycles. “We consume secrets on the fly when we are running the builds, then we delete them,” says Bobba.