At PlaceExchange—a programmatic ad exchange for out of home media—we have architected a system that begins in remote, IoT devices stationed next to freeways or on the interior of subway stations and ends with workers handling messages from asynchronous queues for data ETL processing.

In this session, I’ll share the key decisions we made when provisioning each stage of our infrastructure, from configuring multiple IoT clients on the frontend to the backend systems that ingest and process the data provided. I’ll then dive into some of the constraints triggered as we scaled our infrastructure and elaborate on the lessons learned. I’ll also speak toward how we achieved complete end-to-end observability to not only monitor the overall health of our system in production, but also provide valuable business level metrics for teams beyond engineering in our organization.