Place signaling, SFU, TURN, workers, monitoring, and policy controls in the environment that matches latency, privacy, operational, and commercial requirements.
Run the control and media planes in an organization-managed account, with explicit TLS, ports, firewall, secrets, telemetry, and storage integrations.
Place media, device bridges, workers, or complete RTC stacks close to participants, equipment, and private inference systems.
Signed local license validation and self-hosted deployment support disconnected environments when the license includes an air-gap entitlement and operators maintain an offline renewal process.
TomatoRTC can support architecture, rollout, capacity validation, and incident preparation with explicit responsibility boundaries for any managed path.
Room control, media forwarding, relay traffic, and workflow processing scale as separate operational and economic units.
Measure signaling-only, TURN relay, synthetic SFU RTP, end-to-end browser, soak, and failure scenarios against the target codec, bitrate, fan-out, and geography.
Use the interactive estimator and provider pricing inputs as planning anchors. Internet egress and relay ratios often dominate media cost.
Define health checks, alerting, restart and rollback procedures, certificate and secret rotation, support bundles, and incident responsibilities.
TomatoRTC does not publish a universal “participants per node” number. Target-environment validation is part of the deployment process.
Production auth adapters, tenant isolation, RBAC mapping, waiting rooms, token revocation, admin origin controls, and permission policy hot reload.
WSS, SFU UDP/TCP requirements, TURN relay ranges, TURNS fallback, dynamic credentials, regional providers, and proxy timeout guidance.
OpenTelemetry, Prometheus, usage events, global monitoring, and pluggable customer-owned persistence. In-memory defaults must be replaced where durability matters.
Multi-instance and multi-region abstractions exist, but durable shared registry, production pub/sub adapters, automated capacity routing, and broad failure validation remain foundation work.