Big Ideas in App Architecture

Hosted by Cockroach Labs

BusinessInterviews guests

Website RSS feed LinkedIn Instagram YouTube

Episodes

Latest episode

Aug 2026

Language

About the show

Cockroach Lab’s Big Ideas in App Architecture is a podcast for architects and engineers building modern data-intensive applications and systems. In every weekly episode, an innovator joins the show to share useful insights from their experiences building reliable, scalable, maintainable systems. Welcome to Big Ideas in App Architecture!

Listen to episodes

60 recent

August 19, 202645 min

A Love Letter to the Database: Industry Shifts, Lessons Learned, and What's Next with Perry Krug

Perry Krug has been close to databases his entire career — from Couchbase in the NoSQL era to three years at Pinecone working on vector search. In this conversation with David Joy, the two database industry experts discuss the shifts they’ve witnessed, for better or worse, and why at the end of the day, we will always love the database. David and Perry dive into: The arc from rigid relational systems to NoSQL to today's return of polyglot persistence Vector search and where it shares distributed-systems fundamentals & AI-era production realities Learn more about Cockroach Labs: cockroachlabs.com Learn more about Perry Krug: https://www.linkedin.com/in/perrykrug/

July 22, 202647 min

The Everything Trap: Building AI Software That Lasts with Sam Hilsman

Generating working software with AI has never been easier, but every shortcut a team takes today is a debt it pays back later. In this episode, David talks with Sam Hilsman, co-founder and CEO of CloudFruit , a studio that builds ERPs, analytics platforms, and custom products — including a custom ERP for a Chicago fabrication studio. Drawing on a path that runs from philosophy and healthcare administration into technical leadership, Sam makes the case that speed is the most underpriced risk in AI-assisted building. Sam unpacks his "Everything Trap" thesis: every decision incurs "change debt," so teams should build "icebergs, not ice sheets" — going narrow and deep, tightly scoping, testing relentlessly, and resisting feature bloat. He traces how his own AI use evolved from simple writing tasks to daily coding, and why he now works primarily in Claude Code alongside tools like Playwright for QA and Remotion for video. He shares hard-won lessons from an early venture (HiiBo) around context management and pivoting, the unsolved problem of authentication for agents, and where real moats, governance, and ethical guardrails still need to come from as agents scale.

June 17, 202651 min

Why Inference Engineering Is the Next Big Role in AI with Philip Kiely

Most teams moving AI into production quickly discover that generating an output is the easy part; running it reliably, efficiently, and at scale is a discipline of its own.In this episode, David talks with Philip Kiely, engineer at Baseten and author of Inference Engineering , a free guide to building and operating AI inference systems. Philip argues that every company will soon need a dedicated team to own inference, and draws on his work at Baseten to explain why it demands fluency across GPU optimization, distributed systems, model correctness, and developer experience all at once. He walks through how agentic workloads are reshaping inference demands, which open-weight models are worth watching, and key optimization techniques including KV cache reuse, quantization, and speculative decoding. Grab your copy of Inference Engineering here: https://www.baseten.co/inference-engineering/

May 20, 202645 min

Distributed Systems, Linkerd, and the Cost of Network Calls with William Morgan from Buoyant

In Kubernetes environments, concerns like retries, security, observability, and traffic control often end up scattered across application code, leading to duplicated logic, inconsistent behavior, and systems that are harder to operate and debug. The service mesh was designed to solve exactly that. In this episode, David talks with William Morgan, co-founder and CEO of Buoyant, about the origins of the service mesh and the role Linkerd plays today. Linkerd is an open source, CNCF-graduated service mesh for Kubernetes, built in Rust, that provides security, observability, and reliability. William traces the idea back to his time at Twitter, where the company’s move from a Ruby on Rails monolith to distributed services on Mesos exposed the hidden complexity of network calls and failure handling. He walks through what makes Linkerd unique: operational simplicity, built-in TLS and policy, retries, timeouts, rate limiting, and traffic splitting for safer rollouts. The two also explore open source sustainability, the difference between simple and easy, and how AI may create new platform challenges through new traffic patterns, nondeterministic code, and heightened security needs.

April 15, 202643 min

Making Software as Durable as Data with Peter Kraft from DBOS

Most workflow and orchestration tools rely on external systems and fragile state management, making it hard for applications, especially AI agents to recover from failures or long running interruptions. In this episode, David talks with Peter Kraft, co-founder of DBOS, about a new approach to durable execution. DBOS is a lightweight, database-backed library that checkpoints program state so applications can resume after crashes, upgrades, or API failures. Originating from Stanford research with Matei Zaharia and Michael Stonebraker, the idea is to make software as durable as data by storing application state in the database. Peter walks through what makes DBOS unique: the ability to run in-process while still giving you the durability, distribution, and observability you'd expect from a dedicated orchestration layer. The two also explore how AI agents will create new reliability challenges and why CockroachDB is a natural fit.

March 18, 202650 min

Breaking the Pillars: Rethinking Observability with Charity Majors

Most observability platforms are built around siloed signals, metrics, logs, and traces; not the rich context engineers actually need. The architecture of observability tooling hasn’t kept pace with the architecture of today’s distributed systems. In this episode, David sits down with Charity Majors, co-founder and CTO of Honeycomb.io and co-author of Observability Engineering, to rethink how modern teams understand production systems. Charity argues that the “three pillars” model is outdated, making the case for wide, structured events that preserve context and embrace high-cardinality data so engineers can ask better questions in real time. Drawing on lessons from operating Parse and its challenging acquisition by Facebook, Charity explains how those experiences shaped Honeycomb’s fast, custom-built observability engine. The two also explore how AI agents could soon close the loop between code changes and production validation, and Charity previews the upcoming second edition of Observability Engineering, including a new focus on observability governance for CTOs. Links for Charity: charitydotwtf.substack.com , honeycomb.io/blog , x.com/mipsytipsy

February 18, 202634 min

How to Transform Dev Workflows with CI/CS and AI Agents with Tomer Karin

Most CI/CD pipelines are built to detect failure, not to resolve it. As software systems grow more distributed and complex, that limitation is becoming a bottleneck for resilience. In this episode, David sits down with Tomer Karin, a seasoned software architect in the automotive industry, to explore a new paradigm he calls Continuous Integration and Continuous Solution (CI/CS). Tomer argues that the future of automated software development isn’t just faster feedback loops, it's systems that can autonomously remediate failures using AI. Tomer shares how his years of working on large-scale automotive software revealed the inefficiencies of traditional development pipelines, where engineers spend significant time diagnosing and fixing issues instead of building new capabilities. With CI/CS, AI agents don’t just identify failing builds, they apply fixes that are verified through existing test pipelines, allowing teams to start the next day with a healthier codebase. Join us as we discuss: Why traditional CI/CD breaks down as systems scale and complexity increases What Continuous Integration and Continuous Solution looks like in practice How AI agents can safely diagnose and fix software failures Where human oversight fits into autonomous development pipelines The economic and productivity impact of autonomous software workflows

January 21, 20261 hr 1 min

AI, Market Cycles, and the Systems Built to Outlast Them with Cockroach Labs CEO & Co-founder Spencer Kimball

Most databases are designed for success cases. Real systems fail– the difference is whether they’re built and tested for it. In this episode, David sits down with Spencer Kimball, co-founder and CEO of Cockroach Labs, to explore the architectural and testing decisions behind CockroachDB, and why validating systems under worst-case conditions is essential to building reliable infrastructure at scale. Spencer shares the origin story of CockroachDB, tracing it back to his time at Google, where working on large-scale data systems and helping build Google Spanner exposed the limitations of existing database technologies. Motivated by the limits of existing database designs, Spencer and his co-founders left Google to build a resilient, scalable, open-source database designed for a world where failure isn’t an edge case, it’s the norm. Plus, Spencer and David discuss how emerging AI workloads are reshaping expectations for database infrastructure by increasing scale, stressing latency budgets, and raising the cost of failure. Join us as we discuss: How Google Spanner inspired the creation of CockroachDB Designing a database that assumed failure by default What resilience means at massive scale How AI is changing the demands placed on modern databases How the Cockroach Labs–IBM partnership reflects a broader shift in enterprise modernization Spencer’s futuristic vision for databases, from AI automation to outer space

July 1, 2025Episode 5838 min

How to Scale Data Infrastructure from Startup to Enterprise

In this episode, David sits down with Nishant Raman , a seasoned data infrastructure expert, to explore the evolving world of data engineering, AI integration, and building scalable systems from the ground up. With experience across logistics, healthcare, and fintech, Nishant shares hard-won insights from years of building resilient, cost-conscious infrastructure that scales with company growth and technical complexity. From choosing the right database to navigating the post-LLM explosion in tooling, this conversation offers a rare behind-the-scenes look at what it takes to thrive as the first data hire and beyond. Join us as we discuss: Nishant’s accidental path into data engineering and lessons from startup life The tradeoffs that can come with building scalable, cost-efficient data systems The growing role of AI in data infrastructure and what’s coming next for data teams

May 27, 2025Episode 5740 min

Code, Cloud and Karate: The Unlikely Path to Cloud Architecture

In this episode, David sits down with Masaru Hoshi , a cloud architect at Qlik, to explore his unexpected journey into cloud architecture, the cultural influences that shaped his career, and the evolving disciplines transforming enterprise technology today. Join as we discuss: Masaru’s path from developer to cloud architect and his open-source contributions The rising importance of site reliability engineering (SRE) and finops in modern enterprises The role of AI in reshaping SRE and predictive reliability management What songs should go in a playlist for a Site Reliability Engineer