June 2016 Summaries

LaunchDarkly's product is driven by a REST API, allowing programmatic access to various features such as modifying feature flags and accessing A/B test results. However, there was no way to access raw analytics events due to the large amount of data involved. To address this, an event stream API has been introduced, providing real-time raw analytics data including feature flag requests, analytics events, custom events, and more. This allows customers to use the event stream API for their own internal analytics platforms or build custom integrations tailored to their specific needs. Accessing the API is simple using tools like curl, with documentation available for further information.

The author argues against the use of staging servers, claiming that they increase risk, time, and decrease quality. Many experts agree with this perspective, suggesting that effective continuous delivery and DevOps should not involve staging servers. However, some concerns have been raised about the practicality and validity of eliminating staging servers. The author addresses these concerns by advocating for through and complete automation, continuous integration, feature flagging, and other techniques to ensure smooth transitions from development to production. They argue that staging servers can be replaced with alternative methods such as feature flags, which allow developers to move features to production more quickly and conduct feature validation on the actual production environment.

Edith Harbaugh and Paul Biggar co-host the podcast "To Be Continuous," which covers continuous delivery, software development best practices, and industry case studies. The podcast is part history lesson, comedy show, and educational resource for listeners interested in the software industry. Topics discussed include payment strategies for developers, continuous delivery in mobile development, and the true meaning of SaaS. Notable guests have included Peter Van Hardenberg from Heroku, Kevin Henrikson from Microsoft, and Jocelyn Goldfein, an angel investor.

Feature flags are being integrated into blue-green deployments, providing granular control over feature releases and mitigating risks during software updates. Traditional blue-green deployment methods involved maintaining two production environments, switching traffic between them to ensure stability. However, this approach was more focused on timing and less on testing smaller features. By incorporating feature flags into the process, developers can now manage traffic with a load balancer while gradually rolling out new features to users. This method allows for continuous delivery of small updates while minimizing risks during the release phase.

DevOpsWest, held concurrently with AgileWest, attracted a diverse mix of international and domestic attendees interested in improving their software skills. The event showcased various companies adopting DevOps best practices and continuous integration/continuous delivery methods. One company successfully reduced its code base from 35 million lines to 3.5 million by eliminating unused code, resulting in increased efficiency. Another speaker discussed the importance of responsible use of over-the-air updates for embedded systems controlling critical functions in cars. The event also featured a talk on effective feature flagging for clean deployments and received positive feedback from attendees. DevOpsWest is recommended for those seeking a global perspective on DevOps use cases and patterns.

Cross-platform feature flagging is a technique that allows companies to deliver personalized and synchronized user experiences across different platforms, such as web and mobile applications. By using feature flags, developers can control the visibility of features over time, allowing for real-time personalization, percentage rollouts, and a kill switch for poorly performing features. This approach enables businesses to manage subscription plans by bundling cross-platform features into different tiers, ultimately enhancing user experience and adapting apps to users' needs rather than the other way around.

In this episode of To Be Continuous podcast, Edith and Paul discuss Gartner's Bimodal IT model, which separates software development into two modes - Mode 1 for systems of record and Mode 2 for new projects. They argue against the idea that these two modes can operate independently and emphasize the importance of continuous delivery in mitigating technical debt. The hosts also criticize Gartner's approach to IT, suggesting it is outdated and not applicable to modern software development practices.

Continuous delivery is an era where software development teams use automated testing and QA tools to deliver stable and performant software at accelerated cadences. However, most end-user software releases are coupled with code deployment, relying on testing and QA to identify issues before release. To mitigate risk in feature releases, introducing feature flags (feature toggles) into the continuous delivery process allows for efficient integration of release, deployment, and operational management into the software development cycle. This separation of feature rollout from code deployment enables percentage rollouts, user-targeted releases, and instant feature rollbacks, providing real-world user and performance feedback without the risk of widespread backlash or degraded performance. Companies like Google, LinkedIn, and Facebook have supplemented their continuous delivery processes with feature flagging systems to control feature rollouts, targeted releases, and assess infrastructure performance under new releases.