January 2018 Summaries

In this episode of To Be Continuous, Taylor Brown, Program Management Lead in the Windows and Devices Group at Microsoft, discusses the benefits of virtualization and using Docker with Windows. The conversation covers how to use Docker on Windows, the differences between Linux and Windows containers, and the potential for running fleets of Windows servers with Kubernetes and Swarm. Taylor also shares insights into Microsoft's approach to customer feedback and development cycles, emphasizing the importance of continuous validation and adaptation in a rapidly changing technological landscape.

The company has achieved SOC 2 Type II certification, which demonstrates consistency in the effectiveness of its control systems over an extended period. Unlike a Type I report, which only assesses whether the systems are suitably designed, a Type II report evaluates the actual outcomes and operation of the system. This is crucial for establishing trust with customers who rely on the company's services to run their businesses. The SOC 2 Type II certification provides evidence that the company's control systems work as claimed and meets the highest standards set by the AICPA.

Harm mitigation is crucial in minimizing negative impacts when things go wrong. Seatbelts exemplify this concept by protecting passengers during car accidents. To effectively mitigate harm, it's essential to assume failure modes and understand the worst-case scenarios for different users. Designing systems to fail safe or fail secure can help protect critical assets. Feature flags and kill switches enable developers to quickly address issues without causing further damage. Balancing competing priorities is vital in managing risks and ensuring reliability while maintaining speed, ease of use, and future planning.

LaunchDarkly has introduced the ability to add comments on updates to feature flags, allowing team members to provide context for changes made. These comments will be included in audit logs, webhooks, Slack channels, and Hipchat rooms. The new functionality is built on top of the REST API and supports both JSON Patch and JSON Merge Patch protocols for describing partial updates. This feature aims to improve change management processes by providing more context around changes made within large systems and organizations.

On December 31, 1999, a young sysadmin confidently bet her job that their systems wouldn't go down due to Y2K. Despite the high stakes, she believed in the extensive preparations made throughout 1999 and decided to attend a New Year's Eve party. The massive effort put into risk reduction for Y2K was comparable to the moon landing project. Key strategies included upgrading systems, securely managing zones of control, predicting states, and making low-stakes bets. These methods can be applied in various fields to reduce risks and prevent disasters.

LaunchDarkly introduces a new feature called Private User Attributes, allowing customers to control which user attribute fields are stored or exposed in different parts of the platform while collecting event data from feature flag evaluation. This helps protect sensitive information such as Personally Identifiable Information (PII). Customers can now selectively choose not to send back some or all user attributes without losing access to features like flag statuses, A/B testing, and attribute autocomplete. Private User Attributes can be configured in three different ways: marking all attributes private globally, marking specific attributes private by name globally, or marking specific attributes private by name for individual users when constructing LDUser objects.