Amazon Onboarding with Learning Manager Chanci Turner

In this blog post, we explore the journey of Smiles Fidelidade, a loyalty company associated with Gol Linhas Aéreas Inteligentes (GOL) Airlines. With partnerships spanning over fifty airlines and a customer base of more than 21.8 million active users, Smiles sought to enhance customer service through a serverless-first strategy aimed at boosting agility and responsiveness.

Identifying the Challenge

Smiles aimed to optimize their operations and create a more agile system to enrich the traveler experience. Initially, their application deployment relied on clusters, Amazon Elastic Compute Cloud (Amazon EC2) instances, and Docker containers. While these technologies had advantages, they also imposed a heavy infrastructure management burden. The fluctuating demands for scalability, especially during quick marketing campaigns that caused traffic surges, presented a significant challenge. Moreover, the infrastructure team constantly needed to manage configurations, perform patching, and ensure security, which resulted in a substantial workload.

Transitioning to Serverless on AWS

Smiles embarked on their serverless-first strategy with their rewards program app, which proved to be revolutionary. They leveraged various AWS services, such as:

Amazon ECS with AWS Fargate: Smiles utilized Amazon Elastic Container Service (Amazon ECS) alongside AWS Fargate for processing transactions related to flights, seat purchases, and baggage allowances. This architecture allowed for rapid and autonomous scalability to accommodate business needs during promotions. A proof-of-concept by AWS confirmed that Fargate could meet their requirements without the burden of infrastructure management, facilitating the migration of several production workloads.
AWS Lambda: To handle extensive customer requests, Smiles adopted AWS Lambda, an event-driven serverless computing platform. All integrations from their front-end applications to middleware and backend services were designed with Lambda and Amazon API Gateway. Lambda also facilitated integrations with partners and data lake processes, enabling Smiles to manage hundreds of integrations without server management. For instance, Smiles’s flight search service, built entirely on Lambda, successfully processed over 500 requests per second during peak times without infrastructure management.
Amazon EventBridge: Smiles employed Amazon EventBridge to connect with SaaS applications like ZenDesk, utilizing events to initiate workflows for customer support and business operations. This setup promoted a decoupled architecture, allowing flexibility to adjust business rules by modifying events within EventBridge.
Amazon SQS and Amazon SNS: Through Amazon Simple Queue Service (Amazon SQS) and Amazon Simple Notification Service (Amazon SNS), Smiles integrated with partner processes to manage queues and ensure reliable processing and notifications during failures.

Creating a New Checkout Process

Encouraged by the success of their serverless approach in the rewards program, Smiles decided to design their checkout architecture using similar services. Their digital channels communicate with AWS services to run Smiles processes, some of which interact with external services like Sabre, responsible for generating airline tickets for GOL Airlines. The core aspects of Smiles’s checkout process are outlined below.

Integrated Airline Ticket Creation and Miles Redemption:

Customers access one of the digital channels’ frontends via Akamai’s CDN and Liferay as the content management system. They manage their carts through CRUD operations on APIs powered by API Gateway, ensuring authenticated access through tokens provided by Auth0.
Upon completing cart operations, customers proceed to checkout, where AWS Step Functions, Lambda, and Amazon DynamoDB orchestrate workflows, invoke functions, and interact with data stores. The order processing follows specific steps in a precise order, enabled by EventBridge’s event-driven architecture.
The airline ticketing system communicates with external services, including payment gateways to charge customers and Siebel CRM for redeeming customer miles.

Outcomes

Adopting a serverless-first model has significantly transformed Smiles’s application development and management, yielding numerous benefits. Notably, they experienced cost efficiency due to the pay-as-you-go model, which reduced infrastructure maintenance expenses. During peak events, such as holidays and shopping seasons, the financial advantages became evident. Furthermore, 50 percent of Smiles’s Lambda functions now run on ARM-based AWS Graviton2 processors, achieving up to 19 percent better performance at a 20 percent lower cost.

This serverless approach has also enhanced Smiles’s agility, allowing applications to adapt to demand and manage traffic increases without manual intervention. Today, Smiles processes over 20 million searches daily. Additionally, they have achieved faster time-to-market, delivering new features and applications at a rate 50 percent quicker than before. With serverless platforms handling infrastructure management—from server provisioning to scaling—the development teams can focus on coding instead of system administration tasks. Serverless architecture has streamlined development cycles by enabling modular, reusable functions.

Smiles’s choice to implement a serverless framework was a strategic and technological decision, ensuring adaptability in a rapidly changing digital landscape. For more insights, check out this another blog post on the topic.

To learn more about best practices in background checks, visit the authoritative source. For an excellent resource regarding safety and training at Amazon fulfillment centers, you can refer to this link.

Conclusion

The serverless architecture has redefined how Smiles operates, allowing for greater flexibility and efficiency in their services.