February 2016 Summaries

The talk covers the intersection between distributed systems and distributed development, a new challenge that arises when dividing development into smaller pieces for distributed teams. Understanding this intersection is critical for adopting microservices, which are increasingly being used to deliver better software faster. The speaker aims to address best practices and practical solutions for organizations adopting microservices, while also sharing experiences from companies already addressing the challenges of using microservices effectively. The discussion focuses on how tooling can be developed to enable the adoption of microservices without requiring a custom architecture.

AWS Lambda is a service that provides a way for developers to quickly write services without worrying about the underlying infrastructure. However, upon trying to use it for a real-world HTTP-based microservices project, Ambassador Labs found some uncool things that make Lambda not yet ready for the prime time. These issues include Lambda being a building block rather than a tool, poor documentation, and inadequate error handling. The root of the problem is that AWS provides building blocks like Lambda and API Gateway, leaving it up to developers to connect them manually. This manual process can be tedious and prone to errors, especially when dealing with versioned endpoints or HTTPS. To overcome these issues, various tools are being developed, such as Terraform, Serverless, and Zappa, which aim to automate the process of setting up Lambda and API Gateway. Despite its promise, AWS Lambda is not yet ready for production use due to its limitations in handling errors, providing clear documentation, and offering a seamless user experience.

Creating a microservice requires careful consideration of several key factors before development begins. This includes how it will be tested, configured, and consumed by other parts of the system, as well as its security, scalability, and reliability. Understanding how dependencies will fail and how the rest of the system will handle their failure is also crucial. Additionally, having a plan for upgrading and monitoring the microservice is essential to ensure its continued improvement and reliability. By considering these 10 important questions, developers can create a mature, resilient, and reliable microservice component that meets the needs of their application and minimizes operational complexity.