April 2020 Summaries

Staging environments are criticized for being costly, unreliable, and lacking relevance in real-world scenarios, making them unsuitable for agile teams. Testing in production using feature flags is presented as a more accurate and cost-effective alternative that provides genuine insights into the live environment. The text argues that staging environments are expensive to maintain, often producing unreliable test results that don't replicate those in production, and are generally not prioritized like production environments. Consequently, the focus shifts to testing in production, emphasizing the use of feature flags to safely and efficiently manage feature releases and experiments. The Split Feature Data Platform™ is highlighted as a tool that empowers teams to implement feature flags, conduct A/B testing, and make data-driven decisions without disrupting production.

Harness has introduced several innovations to enhance deployment reliability, including the Pipeline Resume feature, which allows users to continue from failure points, and expanded verification providers through integrations with Instana and Bugsnag. The company also released the Continuous Delivery Insights 2020 report, which provides valuable industry insights and highlights common challenges in Continuous Delivery. In response to global challenges, Harness transitioned to virtual engagements with its community, offering virtual HarnessUs sessions and upcoming webinars to discuss Continuous Delivery topics and trends. Furthermore, the company has expanded its Secrets feature for Cloud Connectors, introduced tagging capabilities for better metadata management, and received media coverage for its contributions to the DevOps field.

Exploring lesser-known features of Java threads, such as thread names, priorities, local storage, daemon threads, and processor affinity, can significantly enhance debugging, performance, and reliability in high-scalability systems. Thread names, which are mutable and can be customized, are particularly useful for debugging by following naming conventions that simplify error-resolution processes. Thread priorities, ranging from 1 to 10, influence CPU time allocation but are merely recommendations for the OS thread scheduler. ThreadLocal storage allows for unique data management per thread, providing valuable context in error handling scenarios, while user and daemon threads determine the criticality of thread completion for process closure. Lastly, processor affinity binds threads to specific CPU cores to optimize cache usage and improve throughput, particularly in high-frequency trading environments, although Java requires external libraries for native affinity support.

Harness streamlines AWS Lambda deployments by using a Harness Delegate installed within an AWS VPC, which connects to AWS services such as S3 buckets and Lambda functions. This setup requires configuring specific IAM roles and policies, including AWSLambdaFullAccess and AmazonS3ReadOnlyAccess. The deployment process involves defining application details within Harness, such as services, target environments, and workflows, with AWS Lambda set as the deployment type. Users can enhance their deployment pipelines with additional steps and monitor their production services using AWS CloudWatch. Harness offers a user-friendly interface for managing serverless functions, allowing for efficient software development and scalability.

The text provides a guide on implementing feature flags in React applications using the Split platform, allowing developers to separate code deployment from feature release for more controlled, gradual rollouts and testing in production. It explains how feature flags function as conditional code to manage feature visibility, and outlines the steps for creating a feature flag within the Split UI, integrating it into a React app, and configuring it to control user permissions, such as enabling or disabling task deletion in a to-do list app. The tutorial highlights the use of the SplitFactory component and withSplitFactory Higher Order Component (HOC) in React to manage feature treatments, and emphasizes the benefits of using feature flags for testing, A/B testing, and canary releases. Additional resources and links for learning about feature flags, integrations, and Split's platform capabilities are also provided, along with the invitation to explore further through a free account, demos, and support options.

Rancher Rio and Harness Platform serve distinct purposes in the realm of Kubernetes and deployment strategies, with Rancher Rio simplifying Kubernetes deployments by focusing on Continuous Deployment and reducing complexity, while Harness Platform excels in orchestrating Continuous Delivery by automating confidence-building steps and facilitating safe deployment strategies like canary releases. Rancher Rio, described as a "Micro PaaS," offers a straightforward installation process and facilitates access to Kubernetes workloads, positioning itself as a deployment engine that requires user input for deployment orchestration. In contrast, Harness provides a comprehensive solution for Continuous Delivery, incorporating various release strategies and automating decisions for safe deployments. While Rancher Rio emphasizes ease of access and deployment within Kubernetes, Harness emphasizes coordinated release strategies and safety mechanisms, making them complementary tools for different stages of the deployment lifecycle.

Implementing Open Policy Agent (OPA) with Harness streamlines authorization modifications across microservices by centralizing policy decisions, which enhances both security and operational efficiency. As distributed systems grow, managing authentication and authorization becomes crucial, often requiring deployments for permission changes. OPA addresses this by decoupling policy decision-making from enforcement, using JSON for data and queries, and Rego for policy logic. The article outlines a step-by-step guide to installing OPA's GateKeeper on a Kubernetes cluster using Harness, emphasizing the simplicity and effectiveness of this integration. With Harness, deploying OPA and adapting to new technologies is made easier, supporting changes at scale. The discussion highlights the partnership and visionary roles of Styra and Harness in advancing OPA methodologies.

Harness has integrated GraphQL into its API, providing developers with more efficient and flexible data queries and mutations, thereby addressing common challenges associated with REST APIs, such as overfetching and underfetching. As an open-source query language, GraphQL allows for declarative data fetching and models business domains in a graph structure, enabling precise data requests and minimizing data transfer between client and server. Harness' API Explorer, which operates as a GraphQL Integrated Development Environment (IDE), facilitates CRUD operations on Harness entities, offering features like schema introspection and strong typing to enhance development efficiency. Users can perform queries and mutations to manage various Harness entities, such as applications, users, and pipelines, with a focus on scoping, hierarchical organization, and future-proofing of APIs. The API Explorer promotes a shared understanding through the use of GraphQL Schema Definition Language (SDL), and its documentation provides examples for constructing queries and mutations.