March 2017 Summaries

NPM, the largest package registry, requires developers to manually publish updates, which can be cumbersome. To streamline this process, the author utilized Codefresh, a platform for automating Docker-based application lifecycles, to automate the publishing of their module. By configuring a pipeline in Codefresh, they automated tasks such as building a Docker image, running tests, and deploying to NPM, triggered by commits to the master branch. This automation not only simplifies the workflow but also enhances efficiency by integrating continuous integration and continuous deployment (CI/CD) capabilities, allowing developers to focus more on development rather than manual publishing tasks. The author encourages developers to explore Codefresh for similar automation needs and invites feedback and discussion on its use.

Before the introduction of Yarn by Facebook in October 2016, NPM dominated as the primary package manager for Node.js, but Yarn quickly gained traction due to its enhancements addressing some of NPM's limitations. Yarn improved upon NPM by offering faster installation times through parallelization and caching, and by ensuring deterministic results in package versioning with its yarn.lock file, as opposed to NPM's shrinkwrap approach which relied on developer maintenance. In continuous integration (CI) environments, both NPM and Yarn face challenges with caching dependencies across multiple servers, a problem that can be partly mitigated by employing Docker's caching capabilities. Codefresh, a platform designed to streamline Docker-based CI flows, offers a solution to these issues by using shared volumes to persist dependencies across different CI flows, thereby enhancing efficiency. To evaluate the performance between NPM and Yarn, Codefresh provides tools and documentation that facilitate testing within its platform.

Kubernetes is a powerful but complex open-source container management system, offering a multitude of features and configuration options that can be daunting for newcomers. In a recent webinar, Raziel Tabib, Co-Founder and CEO of Codefresh, provided a concise introduction to the essential concepts needed to deploy a simple microservices application on Kubernetes in just ten minutes. The webinar highlighted the importance of Kubernetes entities like Pods, Deployments, and Services in managing containerized applications. Deployments simplify the management of Pods by automating tasks such as creation, scaling, and health monitoring, while Services facilitate internal and external communication by providing a consistent IP and port. The live demonstration included in the webinar illustrated these concepts by showing how Deployments can roll out updates efficiently and maintain service continuity even if a Pod fails or is updated.

The recorded webinar, co-hosted by Codefresh and Cloud66, offers insights into testing strategies and best practices for Docker containers, focusing on their deployment in production environments. Alexei Ledenev from Codefresh presents four testing strategies, showcasing their benefits and drawbacks through demos, while emphasizing the importance of creating portable and efficient Docker environments. The session also features a talk from Daniël, who guides participants through the stages of integrating containers into production workflows, highlighting the need for a structured approach and the "artichoke model" for embracing change. Alexei advocates for a Docker Automation Flow, which involves creating separate builder, test, and app containers to enhance portability and streamline the build process. Despite the complexity of setting up such a flow, tools like Habitus.io and Docker CI/CD services are recommended to simplify automation and orchestration. Participants are encouraged to sign up for accounts with Codefresh and Cloud66 for further support and exploration of Docker-driven development.

Codefresh’s Docker CI service has enhanced its build speed capabilities by introducing full layer caching, a feature previously exclusive to paid accounts, to free accounts as well. This improvement is made possible through Docker 1.13, which allows for the sharing of caches among build servers. As a result, when building an image for the first time, all layers are pulled and cached, and subsequent builds automatically utilize this cache without additional configuration. This automatic caching sets a previous image as the base, eliminating the need to pull layers again, which significantly boosts build speeds. While the specific speed enhancement varies with different applications, benchmarks indicate a typical increase of 5-6 times faster builds, offering users a quicker and more efficient development process.