FastCGI microcaching is a clean single-box pattern for dynamic pages that barely change second-to-second. The ops question grows when you add more edges: who owns purge of those tiny TTLs, which nodes hold which sites, and how you verify a purge actually landed. We have been building an open-source CDN control plane (CDNCA, AGPL-3.0) around self-hosted nodes and a plugin marketplace; marketplace features roll out release by release. Did you keep the microcache TTL under a second in production, or did you stretch it once the hit ratio looked good?