May 2019 Summaries

Harness's Continuous Delivery platform strengthens operational integrity and compliance by providing comprehensive audit trails that document every change in deployment pipelines, capturing details such as timestamps, event sources, and resource updates. These electronic records are crucial for maintaining transparency and accountability, particularly in answering critical questions like "Who did what when?" within large enterprises with diverse applications and tech stacks. The audit trail feature, accessible under the Continuous Security navigation, logs user-level transactions, such as creating, updating, or deleting resources, and persists this information for a year. It also highlights detailed differences for updates, thereby supporting organizations, especially in financial and retail sectors, in proving compliance and maintaining operational integrity.

Harness has implemented ephemeral environments in Kubernetes to enhance deployment flexibility and team collaboration by enabling developers to deploy feature branches for testing without bottlenecks. This approach facilitates stress testing, upgrade path verification, and data state preservation across various testing scenarios, moving beyond initial constraints of limited deployment environments. By allowing the creation of temporary environments that can be set up and dismantled as easily as a git branch, Harness offers developers a scalable and cost-effective solution to collaborate across different teams, including Quality Engineering, Security Operations, and UX Design. Utilizing Kubernetes ingress and Harness, these environments can be created on-demand and used for testing different resource configurations, migration paths, and data states. This setup ensures that testing can be performed in isolated environments, reducing the competition for resources and providing significant flexibility for developers. The process involves configuring namespaces with specific metadata to manage resource usage effectively, ensuring environments are automatically dismantled after their designated time, thereby optimizing resource utilization.

Harness has integrated with PagerDuty to enhance incident management and improve DevOps team responsiveness during deployment failures by providing real-time deployment notifications. PagerDuty, which has become a standard tool for alerting developers about incidents, recently went public with over 11,000 customers. Harness aims to empower developers with self-service deployment capabilities, making continuous delivery the norm. The integration allows users to create notification strategies that alert specific user groups based on various execution conditions, such as deployment successes, failures, or pauses. Setting up the integration is straightforward and involves editing notification settings within Harness and using a PagerDuty integration key. This development is part of Harness's effort to support requested notification channels like email, HipChat, and Slack, alongside PagerDuty.

Dynamic Configurations in the Split platform enable teams to modify UI elements, logic, and performance settings without requiring engineering changes, allowing for faster experimentation and optimization of feature variations. This functionality helps teams quickly adapt to user feedback by providing a centralized interface where configurations can be adjusted, such as UI colors, educational copy, and algorithm weighting, all without direct code intervention. Additionally, the platform supports complex configurations through a JSON editor, facilitating fine-tuning of machine threading for performance optimization and connection handling in distributed architectures. By leveraging these capabilities, teams can conduct feature experiments and manage feature flags more efficiently, enhancing the speed and effectiveness of product development and deployment.

Continuous Integration (CI) and Continuous Delivery (CD) are distinct processes within the software development lifecycle, with CI focusing on transforming code into artifacts and CD emphasizing the safe, quick deployment of these artifacts into production. Despite the crowded market of over 35 vendors claiming to offer CI/CD solutions, true CD practices, characterized by frequent production deployments, remain rare, with companies like Harness standing out. Historically, software release processes involved extensive project management and coordination, but the advent of Continuous Delivery, cloud computing, and DevOps has revolutionized these practices, promoting automation and increased developer velocity. DevOps, often confused with Continuous Delivery, is more about fostering a culture of collaboration, automation, and shared goals, while CD is a practice that can thrive within such a cultural environment. The article highlights the evolution of these concepts and the limited number of vendors genuinely practicing Continuous Delivery, emphasizing the importance of distinguishing between CI/CD tools and their roles.

Harness, in conjunction with Stackdriver, streamlines and accelerates the process of verifying deployments by leveraging automation and machine learning, significantly reducing the time and resources required for deployment verification. This integration allows companies like Build.com to cut down the manual verification time from several hours to just 15 minutes per deployment with the help of a single engineer. Stackdriver, a monitoring tool acquired by Google, provides comprehensive metrics, logs, and event data from cloud infrastructures, which Harness analyzes using unsupervised machine learning to identify anomalies or regressions in deployments and trigger automatic rollbacks if necessary. By automating canary deployments and health checks, companies can increase deployment frequency and focus more on development rather than manual verification, supporting a more efficient Continuous Delivery process. The integration is seamless as Harness automatically discovers the Stackdriver APIs for registered cloud accounts, and users can configure deployments to incorporate Stackdriver verifications in various deployment workflows, ensuring all metrics are validated and the deployment operates within normal parameters without requiring extensive manual oversight.

Harness enhances the capabilities of Helm, a popular tool for managing Kubernetes applications, by introducing advanced deployment strategies like Canary and Blue-Green deployments and automating rollbacks. This integration allows for streamlined Kubernetes application management without altering the chart specifications, offering a first-class continuous delivery solution for microservices packaged as Helm Charts. Harness simplifies environment-specific configurations and provides automatic rollback capabilities, addressing the manual work typically associated with Helm's deployment and rollback processes. By organizing environment-specific configurations and secrets within an Environment resource, Harness facilitates the deployment of services across multiple environments. The platform also supports Helm Repository Connectors and allows Helm charts to be fetched directly from Git repositories, enabling a GitOps workflow. Additionally, Harness supports client-only mode for Helm, circumventing the challenges associated with tiller and leveraging standard kubectl mechanisms for deployment.