Schedule: Case Study sessions

You have a large talented group of engineers, each with an opinion on the best programming language, build tool, test framework, and CI/CD pipeline. How do you get them to agree and reuse without demotivating them? Andrey Utis explains how Capital One mandated a single CI/CD framework and kept users engaged through a contribution process, empathy interviews, and openness. Read more.

When personalizing for news platforms, you must pay attention to both the audience and the content. Much like most software architecture systems, readers and news content should not be treated as a monolith. Katerina Iliakopoulou details the architecture of a system that facilitates agile personalization on the New York Times’s platforms to accommodate fast A/B testing. Read more.

So you think you can test your complex distributed application effectively just using your test environment? At Netflix, automated testing of client and server applications runs at scale in production. It has quickly gone from low-volume manual mode to automated continuous and voluminous mode. Vasanth Asokan offers a study of such testing at scale that will inform your overall testing strategy. Read more.

Threat modeling as a discipline has always enjoyed a special place in development, going from "Why do it?" to "I should do it one of these days" to "We did it and didn't even get a T-shirt." Many competing methodologies, interests, and constraints help make the process more difficult than it needs to be, reducing the results. Izar Tarandach shares the approach Autodesk uses for threat modeling. Read more.

Ryan Cooke explains how one organization transformed its architecture from a few hundred disparate servers to running entirely in container orchestration within six months. While the project met its goal to reduce infrastructure costs, it also realized an unexpected benefit: greatly improving developers' velocity in delivering software into production. Read more.

Despite its success in building an engineering and DevOps culture, when Capital One embarked on the gargantuan task of embracing microservices several years ago, it had to do an analysis of which of its development and operational practices were applicable to the new architectural style and which had to be fundamentally reevaluated. Irakli Nadareishvili shares lessons learned from the process. Read more.

Have you ever launched a large ETL job to check a fix for a corner case in a derived calculation or normalization? Marc Siegel shares lessons learned applying the event sourcing pattern within an ETL pipeline. Key takeaway in regex form: E{1}T*L* -- that is, "Extract once, transform and load infinite times." Read more.

How do you develop great architects on your team? It's a real challenge when your organization offers limited opportunities to actually perform as an architect. Jean Bordelon shares approaches to give aspiring architects meaningful ways to grow and veteran architects ways to hone their craft, as well as lessons learned along the way. Read more.

While the rest of the software architecture world is admiring their containers, edge computing, and cloud-native architecture, game developers are off in the corner creating entity component system (ECS)-based architectures and pushing the boundaries with this flexible, compatible, composable approach. It's not just for games. Paris Buttfield-Addison, Mars Geldard, and Tim Nugent explain why. Read more.

Audible.com delivers millions of hours of audio content daily across a range of mobile apps and devices. As the company has grown, it has had to quickly scale to meet demand and provide the best experience for its customers. Tyler Lund explains how Audible's audio delivery and playback architectures have evolved and how the company utilizes chaos engineering at scale to improve reliability. Read more.

How can a large software organization strike a balance between gaining leverage from the use of common technologies and empowering teams to make their own decisions? Jon Moore outlines the Architecture Guild framework Comcast uses to try to thread this needle. Read more.

How do you bring safety back into an organizational culture when the contemporary patterns used to increase the rate of change also contribute to increased fragility? Subbu Allamaraju examines contributing factors, the limits of chaos testing, and patterns and practices needed to support a high rate of change while also maintaining system safety. Read more.

Vladik Khononov explains how he and his team embraced domain-driven design (DDD) at Plexop, a large-scale marketing system that spans over a dozen different business domains. Join in to learn how DDD allowed the team to manage business complexities, see what worked (and what didn't), and discover where they had to adapt the DDD methodology to fit the company's needs. Read more.

Comcast is evolving a cloud-based data infrastructure to support classic analytic use cases, real-time operational analysis, and modern machine learning. All these use cases require finding high-quality data of interest, understanding its semantics, and tracing its route from streaming ingestion to at-rest storage. Barbara Eckman explains how a judicious data governance strategy fills these needs. Read more.

Matt Chapman leads a walkthrough of the architecture and open source components that serve Tribune Publishing's content recommendation system, powered by online machine learning at scale. Find out how multiple publications, multiple recommendation algorithms, and one scalable architecture regularly achieve double the performance of the legacy solution. Read more.