August 2018 Summaries

The author was surprised by the wide range of use cases supported by LaunchDarkly's feature management platform. Initially, they thought it was a tool for software development, but after hearing about their friend's small software company facing challenges in satisfying customers who demanded unique features, they realized that feature flags could be used to provide differentiated experiences and custom pricing structures. This solution enables vendors to capture appropriate value while users pay only for what they need. The discovery of novel applications of the feature flagging approach will continue as more problems are solved using this fundamental technique.

Feature flags are crucial for managing application features and can be temporary or permanent. Progressive Delivery allows exposing flagged features to more users over time. To provide better visibility into flag changes, enhanced audit log entries have been released, showing details of rule additions, deletions, reordering, value changes, multivariate flag rollouts, and more. These enhancements are now available in the audit logs and for LaunchDarkly/Slack integration users.

LaunchDarkly has added support for metrics and traces in its Relay Proxy (LD-relay) to provide better visibility into service metrics. Customers can now forward these metrics to standard monitoring tools like DataDog, StackDriver, and Prometheus. To use this feature, users need to create a configuration file specific to their chosen monitoring application and ensure that the settings in the monitoring tool are ready to consume data. The supported metrics include the number of requests received, open streaming connections, and route traces, which provide insights into the performance of the relay proxy.

Software delivery has evolved to encompass not just Continuous Delivery, but also Incremental and Progressive Delivery. Continuous Delivery ensures code is always in a deployable state by increasing the frequency of commits, builds, tests, and deployment. This reduces risk as changes are smaller and problems are identified more quickly. Incremental Delivery involves breaking large projects into small units that can be released to customers as soon as they're complete, allowing for faster feedback and adjustments. Progressive Delivery delivers changes first to low-risk audiences before expanding to larger groups, reducing the risk of each change by controlling exposure. Achieving all three practices can help teams move faster with less risk.

Dequan Zhang, an undergraduate student and LaunchDarkly intern, shares his experience in the ScriptEd program, which teaches students how to code in HTML, CSS, and JavaScript. After joining the program in his junior year of high school, he was selected for a summer internship at LaunchDarkly where he worked on improving their internal administration application called Vitalstatistix using feature flags. Despite challenges such as learning ReactJs on-the-job, Dequan gained valuable experience and insights into the tech industry, including meeting with the CEO and CTO of LaunchDarkly. He encourages future ScriptEd interns to embrace the opportunity and take advantage of networking opportunities in the industry.

Jian Xie, an undergraduate student and LaunchDarkly intern, shares their experience in a guest post. They preferred working on a small engineering team as it allowed for better relationships with peers and faster project completion. Their favorite part was interacting with the people at LaunchDarkly, who were welcoming and interactive. Initially, they found learning a new language challenging but improved over time with help from the team. The internship involved working on an internal administration dashboard and displaying server connection data in charts for valuable insights. Jian also highlights the importance of making connections and gaining exposure to different people in the industry through internships.

Software development has evolved over time, mirroring changes in delivery mechanics. Early stages involved physical hardware devices for software procurement, leading to long cycles that mapped to hardware design and validation intervals. Waterfall complemented this staged process well. As packaged software decoupled from hardware procurement, teams sought faster ways to deliver software, leveraging the validation stage for beta testing or early access programs. Agile and Scrum models emerged with the idea of breaking down movement towards a goal into small changeable tasks or stories. This allowed minor course corrections on the journey towards the big picture. The advent of Software as a Service (SaaS) led to continuous delivery, where updates could be made at any time without physical deliveries. However, this also meant immediate exposure to bugs or errors. Feature flagging was used to separate code deployment from feature release, allowing developers to make minor changes and fix bugs without redeploying the entire application. Progressive Delivery emerged as a new approach that staged the release of features or updates in a way that managed their impact. This involved separating code deployment from feature release and controlling the release like a valve or gate that could be slowly opened and closed. It also required feedback data about how various features were being accessed and consumed. While Continuous Delivery can be done with or without a feature management platform, Progressive Delivery requires this aspect of control. This approach incorporates a 'built-for-failure' model through the use of feature flags and a feature management platform that consolidates all control points into a single interface for organizations of any scale.