As geospatial applications increasingly move into cloud environments, the traditional monolithic approach to serving feature data is starting to show its limitations. Teams often encounter familiar challenges such as scaling bottlenecks, infrastructure lock-in, and increasing friction when integrating geospatial services with modern distributed databases and cloud platforms.

This talk explores a practical approach to rethinking how feature data services can be designed and operated. By adopting a composable architecture, we decouple the API layer from the underlying database storage, allowing each component to evolve and scale independently. This architectural shift enables geospatial APIs—such as OGC API - Features—to become significantly more flexible, deployable as lightweight services, and adaptable to a variety of modern data stores.

To demonstrate this approach, we introduce Meros, an open-source feature data service developed by our team. Meros implements a modular architecture that separates the API layer, feature service logic, and database adapters. This allows the system to seamlessly integrate with different databases while remaining simple to deploy and operate.

In this session, we will share the journey of designing and building Meros from the ground up. We will explore the key architectural decisions behind the system, discuss the operational benefits of a decoupled design, and present practical considerations for deploying and scaling its components. By the end of the talk, developers and architects will walk away with a practical case study—giving them a tangible starting point to build their own resilient, adaptable feature data services.