July 2019 Summaries

In this article, Erez Rokah compares CircleCI with AWS CodeBuild for serverless applications and focuses on the advantages of using CircleCI. CircleCI offers several key benefits, including free private repository access, a mature and well-documented tool, integration with other development tools, relevant notifications, a REST API to manage build jobs, and a reproducible setup that allows adding new services with ease. In contrast to CodeBuild, CircleCI does not use CloudFormation to describe the deployment process, has no regions, requires giving CircleCI access to the AWS account, and handles email notifications and GitHub access tokens automatically. Erez Rokah sets up an advanced CICD process using CircleCI and provides a CLI tool that follows a GitHub project, connects it to AWS, and sets up environment variables and project settings. The setup includes different deployment stages, automated tests, and the ability to deploy specific services with their dependencies.

Amazon has announced EventBridge, a rebranded version of CloudWatch Events with new features. The service is designed to deliver system and user events in real-time, like an event bus, but was previously badly marketed by AWS. It can capture user-published events as well as system events from other AWS services and deliver them to various targets, including large enterprises that want to centralize their application governance. EventBridge now supports custom events from AWS's partners such as DataDog and OneLogin, reducing the need for webhooks and allowing users to handle these events using Lambda functions or forwarding them to supported targets. Users can also augment partner events with custom user events from their application, creating a complete audit trail of all events that occurred. EventBridge brings partner events closer to the user's application, taking care of event ingestion so they can focus on adding value to their business.

The author reflects on their experience with a greenfield project in 2016 where they implemented new technologies such as serverless, event sourcing, and functional programming. However, they soon realized that the increased complexity of these new tools led to an excessive amount of "glue code" - repetitive tasks needed to connect different components together. The author believes that this trend is on the rise due to the increasing popularity of smaller and more specialized components, which leads to a higher surface area between components. This results in a decrease in productivity and an increase in maintenance costs. The author proposes the term "surface area" to describe these tasks and suggests that reducing the cost of surface area should be a priority over creating larger components. They also mention that serverless technologies can help alleviate this problem, but more work is needed to make it easier to manage surface area.

This article discusses the development workflow of a 100% serverless company, Lumigo. The guiding principles and tools used at Lumigo aim to ensure an efficient workflow. Serverless development requires a different approach, focusing on adopting a mindset that prioritizes efficiency and embracing managed services. Key aspects of the workflow include detecting problems early, automating testing and linting, and using cloud-based toolchains. The article also highlights the importance of developers being responsible for monitoring and maintaining their code. The technical stack used at Lumigo includes Python 3.7, NodeJS 8.10, Jira, GitHub, Pytest, Flake8, Serverless, CircleCI, CodeCov, and lumigo.io. The workflow is designed to be flexible and adaptable, with a focus on continuous integration and delivery.