January 2021 Summaries

Progressive Delivery is a strategic approach to software deployment that emphasizes learning from failures and minimizing their impact by delivering changes incrementally to small, low-risk audiences before gradually expanding to larger groups. This method allows for real-time validation and feedback, enabling companies like Target and IBM to manage feature releases more effectively. Key components include release progression, which involves carefully selecting the audience to limit risks, and delegation, which empowers relevant teams to manage features and respond swiftly to issues. Utilizing techniques such as A/B testing, chaos engineering, and canary testing, Progressive Delivery ensures a controlled and adaptable rollout process. Essential tools like feature flags and observability help identify and recover from failures, allowing for targeted action such as load shedding or request throttling. By fostering a culture that anticipates and learns from failures, Progressive Delivery equips organizations to deploy software safely and efficiently.

In 2020, the focus of DevOps shifted to people and processes as companies adapted to remote work during the pandemic. Remote work brought challenges such as working asynchronously, managing async communication, maintaining productivity, and adjusting to rapid pivots in business strategies. The disappearance of network perimeters forced IT teams to adopt least-privilege security practices and support full connection to work resources from various endpoints. Companies also had to do more with less due to cost-consciousness, leading to investments in tools and solutions for the distributed workforce. Software played a crucial role in helping organizations adapt quickly during these challenging times.

IBM has been using feature flags from LaunchDarkly for several years, significantly improving their deployment frequency and enhancing control over product delivery. The use of feature flags as a kill switch has allowed IBM to block bad actors from abusing the free tier of its cloud service, saving time and money while reducing opportunity costs. Additionally, LaunchDarkly's feature flags have facilitated progressive rollouts for new data centers, virtually eliminating risks associated with software releases. Overall, LaunchDarkly has contributed to increased developer productivity, reduced development costs, and improved system performance at IBM Cloud.

Trunk-based development is a version control branch management method that aims to reduce complexity and user error by eliminating long-lived feature branches in favor of developers merging their code into a single "trunk" branch. This approach helps teams avoid merge conflicts, increase deployment speed, and improve overall productivity. Key benefits include short-lived branches, fewer merge conflicts, quicker releases through continuous integration and delivery (CI/CD), and the need for an extensive automated testing scheme and feature flags to manage incomplete features.

LaunchDarkly, a feature management platform, now serves peak days of 20 trillion features. Founded in 2014 by John Kodumal and Edith Harbaugh with the goal to help software teams move faster with less risk, LaunchDarkly has seen significant growth and adoption among various industries. The company's customers include Recreation.gov, Seesaw, and Intuit, who use feature flags for controlled rollouts, managing technical debt, and enabling product teams to run powerful use cases. In 2021, LaunchDarkly plans to host its first user conference, Galaxy, on April 6th-7th, and is actively hiring in various positions across their Oakland, Atlanta, and UK hubs.