Databases and Frameworks at the Edge with Glauber Costa and Igor Minar
Published:
A conversation with Glauber Costa and Igor Minar on how databases, frameworks, and edge computing shape modern web development.
Episode Description
A conversation with Glauber Costa and Igor Minar on how databases, frameworks, and edge computing shape modern web development.
Episode Summary
In this episode, host Anthony Campolo sits down with Glauber Costa from Turso and Igor Minar from Cloudflare for a detailed exploration of how databases and frameworks can thrive at the edge. Glauber shares his journey from contributing to the Linux kernel and building ScyllaDB to creating Turso, a service powered by a fork of SQLite that brings data closer to users. Igor discusses his experiences transitioning from leading the Angular project at Google to working on Cloudflare’s developer platform, emphasizing how global edge networks can reduce latency and improve performance. Together, they address key topics such as the complexities of data distribution, the practicalities of serverless and edge deployments, and the evolving role of front-end developers in database decision-making. By the end, listeners gain a clearer sense of the real-world benefits and potential pitfalls of pushing both compute and data to the network edge, making for an insightful discussion relevant to anyone building modern web applications.
Chapters
00:00 - 05:03 - An Unconventional Start & Open Source Beginnings
This section sets the stage with host Anthony Campolo introducing the live podcast setting and welcoming Glauber Costa. Glauber reflects on his early career decisions, which went against conventional wisdom at the time—he dove headlong into Linux kernel development when peers warned there was “no money” in open source. Drawing parallels to industry uncertainties like the 2001 tech bubble, Glauber highlights how passion for a technology stack can lead to unexpected opportunities. He then recalls how he joined Red Hat, exploring the once-radical idea of making software free and charging for support. These formative stories illustrate how open source models disrupt traditional job prospects and pave new paths for innovation. They also foreshadow the later conversation about bridging core system-level work (operating systems, kernels, databases) with emerging front-end and edge technologies to produce more efficient developer experiences.
05:03 - 10:05 - Shifting to Databases & The Birth of Scylla
As the conversation continues, Glauber recounts his transition from kernel engineering to tackling a startup that initially tried unikernels—lightweight, single-purpose operating systems. When that approach struggled to find traction, the company pivoted toward reimplementing Apache Cassandra in C++, leading to ScyllaDB. He describes how Scylla utilized techniques from kernel development to improve performance, such as direct I/O and avoiding memory-mapped files. This yielded a database often performing at multiples of Cassandra’s speed, drawing new users and enterprise customers. Glauber explains the shift in how modern teams choose databases, noting that infrastructure or DevOps engineers frequently make key data decisions instead of traditional DBAs. This chapter culminates with a look at how broad industry trends—containers, distributed computing, and varied application needs—influenced the evolution of next-generation databases.
10:05 - 15:02 - From ChiselStrike to Turso: Forking SQLite for the Edge
Glauber shares the founding of ChiselStrike, a project aimed at bundling a database with an API layer to simplify backend tasks for front-end developers. He recalls the challenges of merging TypeScript definitions with an operational engine, ultimately recognizing that the embedded SQLite portion garnered the most interest. That realization drove a company pivot to Turso, focusing primarily on forking SQLite to suit distributed edge deployments. Glauber explains the technical and licensing nuances of SQLite, including its public domain status and closed-contribution development model. By forking the code into “LeapSQL,” Turso can add features like real-time change streams and WebAssembly user-defined functions, while also welcoming community contributions. This chapter highlights how developer feedback can pivot entire products toward simpler, more flexible solutions—and how bringing compute and data closer to the user can unlock new performance possibilities.
15:02 - 20:08 - The Economics of Edge & Why SQLite Fits
Here, the conversation focuses on why an embedded engine like SQLite is uniquely suited for edge computing. Glauber outlines how spinning up massive data centers in dozens of global locations is cost-prohibitive unless the database footprint is extremely small and resource-efficient. SQLite’s minimal overhead and direct file-based approach make it easier to replicate across many locations, drastically reducing latency. The group underscores that moving from one or two regions to 30 or more demands a different mindset than traditional cloud deployment. This shift, described as the true essence of “the edge,” allows developers to stay unaware of the underlying infrastructure’s geographical complexity. As an example, Turso’s simple URL-based system routes queries to the nearest replica, showcasing how platform abstractions can eliminate the manual overhead of multi-region orchestration.
20:08 - 25:13 - Onboarding Developers & Building with Turso
With the business rationale established, Glauber details how developers can get started using Turso’s free tier, which supports three global locations and large read allowances. The conversation touches on the importance of offering generous tiers that enable real experimentation without hidden costs. The group also discusses the significance of developer community engagement—Discord channels, open contribution projects, and bridging the gap between low-level performance tuning and high-level front-end demands. The hosts emphasize that while databases can seem daunting, the right abstractions enable front-end or full-stack developers to handle everything from migrations to distributed caching. This section provides concrete advice on how to begin exploring data at the edge with minimal friction, exemplifying how Turso aims to democratize specialized database knowledge.
25:13 - 30:10 - Enter Igor Minar: From Angular to Cloudflare
In this chapter, Igor Minar joins the discussion. He introduces his background as a decade-long lead on the Angular project at Google, explaining how that experience laid the groundwork for a broader interest in unifying front-end and back-end layers. Now at Cloudflare, Igor focuses on developer experience within a company traditionally seen as a CDN and security provider. He highlights Cloudflare’s robust global network—with hundreds of data centers worldwide—and reveals how that capacity is leveraged for solutions like Cloudflare Workers, R2 storage, and D1 edge databases. By drawing on parallels with Angular’s open contribution model, Igor underscores the importance of community feedback in guiding product direction. His presence shifts the conversation to a more holistic perspective on how frameworks, servers, and data distribution must all cooperate to deliver optimal user experiences.
30:10 - 35:00 - Data Placement, Smart Routing & the Reality of Latency
Igor expands on the complexities of server-side rendering at the edge, cautioning that blindly moving compute away from data can produce worse performance if data remains in a single remote region. He discusses how Cloudflare’s “smart placement” analyzes traffic and determines whether to run an application near the user or near the data, effectively bridging cloud and edge environments. Glauber echoes these insights, reiterating Turso’s focus on replicating data globally to avoid such bottlenecks. The duo acknowledges that many organizations are locked into existing data agreements (with AWS, Oracle, or on-premise hardware). This section underscores that true edge success often demands either migrating data closer to users or intelligently routing compute where the data resides, preventing the pitfalls of ill-considered deployments.
35:00 - 40:48 - Converging on Edge Best Practices & Closing Thoughts
In the final chapter, Anthony steers the conversation toward best practices for building high-performance applications that blend front-end frameworks and globally distributed data. Igor and Glauber highlight the importance of embracing platform intelligence—whether via Cloudflare’s smart routing or Turso’s multi-replica model—to ensure low latency. They also emphasize the collaborative nature of these technologies: while Cloudflare’s D1 might seem a competitor to Turso, both exist to help developers solve data bottlenecks in different ways. The episode wraps up with contact information, guiding listeners to Turso’s website, Cloudflare’s developer resources, and personal Twitter handles. Reflecting on the shared mission to improve developer experience at the edge, the guests leave listeners with a final note of optimism about the continuing evolution of serverless frameworks, open source communities, and distributed databases.