Submitted by: TrellisTech ( Shambu Pujar and Marek Karwacki) Date: February 2026 Version: 1.0 Infrastructure: Amazon Web Services (eu-west-1) / OVH UK (Option B) Related Proposal: Babylon Gateway Operator Proposal
Following our submission of the Babylon Gateway Operator Proposal (AWS Edition), TrellisTech team proposes to additionally operate two lightweight Radix ecosystem services: the Radix Connect Relay (RCR) and the WebRTC Signalling Server. Both services function as untrusted message relays — they facilitate encrypted communication between Radix clients without inspecting, storing, or logging message payloads.
These services are architecturally simple: stateless application pods backed by ephemeral Redis storage. Neither service requires a persistent database, long-running data synchronisation, or complex multi-component coordination. This makes them ideal candidates for deployment as lightweight additions to existing infrastructure.
This proposal presents two deployment options:
| Aspect | Option A — K8s Add-on | Option B — Standalone (OVH UK) |
|---|---|---|
| Prerequisite | Accepted Gateway proposal (shared EKS cluster) | None — fully independent |
| Monthly fee | $1000 | $2,500 |
| Deployment time | 1 week | 2–3 weeks |
| Achievable SLA | 99.9% | 99.0–99.5% |
| Monitoring | Inherited (Prometheus / Grafana) | Must build separately |
| Security | Inherited (WAF, Shield, Istio, VPC) | Local firewall, nginx |
| Operational overhead | Minimal — add pods to existing cluster | Significant — new servers, TLS, monitoring, legal entity |
| Recommendation | Recommended | Fallback if Gateway proposal not accepted |
Recommendation: Option A is strongly preferred. The combined resource footprint of both services is negligible on the existing EKS cluster. Option A eliminates redundant infrastructure, and inherits the full security, monitoring, and high-availability stack already built for the Gateway.
Key commitments:
| Metric | RCR Target | RCR RFP | Signalling Target | Signalling RFP |
|---|---|---|---|---|
| SLA / Uptime | 99.9% | 99.9% | 99.9% | 99.9% |
| Global latency (TCP) | < 300 ms | < 300 ms | < 300 ms | < 300 ms |
| Redis GET/SET latency | < 1,000 ms | < 1,000 ms | < 1,000 ms | < 1,000 ms |
| Concurrent connections | N/A | N/A | 6,000+ | 6,000 |
Soft SLA: All performance and availability targets are best-effort commitments rather than hard contractual guarantees. Both services are public and unauthenticated, exposing them to intentional abuse and unintentional overload. The operator will make every reasonable effort to meet or exceed the stated targets and will be transparent about any shortfall and its cause.
This is an infrastructure operations proposal. The scope is explicitly limited to deploying, operating, and maintaining the Radix Connect Relay and WebRTC Signalling Server software as released by the Radix community and RAC approved versions.
In scope:
Out of scope:
Continuity assumption: This service is viable for as long as official releases can be deployed without custom code changes. If a situation arises requiring application code modifications, the operator will notify RAC as early as possible and maintain the current version in service during any transition period.
SLA exclusion: Any downtime or degradation arising directly from a condition outside the operator’s scope — including application-layer bugs, protocol changes requiring code fixes, or abuse beyond normal operational bounds — is excluded from SLA calculations.