Vector tiles have become the primary method for dissemination of OpenStreetMap data. The performance of a vector tile-based map largely depends on the size and complexity of individual tiles and the features they contain. While these chunks of data are spatially homogenous, they are heterogenous in content, much like the landscape they represent.
How do we understand tile performance across this varied landscape? In this talk we’ll outline our preferred methods for understanding the weight and variation of data in vector tiles — using techniques like bash pipes, tile linting, and golden ratio sampling — and make the case for optimizing your tile sets through cartography-driven tile engineering.
No recording is available from this session.