What this node doesthree workloads, one edge
Selective forwarding
Small conferences (3–12 participants) are forwarded by an SFU: each participant sends one DTLS-SRTP stream and receives one per peer. Media runs over UDP and is never re-encoded.
Segmented delivery
Recordings and live segments are served over QUIC on UDP/443. Loss recovery per stream keeps playback smooth on mobile networks where TCP would stall.
Session ingest
Clients ship quality metrics (RTT, jitter, loss, bitrate) in batched QUIC datagrams so we can move a struggling participant to a lower layer within a second.
Traffic profileas published for peering and capacity planning
| Dominant protocol | QUIC v1 on UDP/443 (HTTP/3 and WebRTC media), plus DTLS-SRTP on
UDP 1024–65535 |
|---|---|
| Sustained egress | 20–120 Mbit/s; bursts to ~800 Mbit/s during scheduled sessions |
| Typical packet size | 1200–1400 bytes (QUIC avoids IP fragmentation; Path MTU is discovered with DPLPMTUD) |
| Concurrency | a few hundred QUIC connections, each long-lived (minutes to hours) |
| Congestion control | BBR on both the client and the relay path |
| Encryption | TLS 1.3 inside QUIC; DTLS 1.2 for media. No plaintext on the wire, no CDN in front |
| Why UDP | Real-time media cannot wait on TCP retransmission order. A single lost packet must not block the frames behind it — loss is handled per stream, above the transport. |
If you are investigating traffic from this address: it is expected to be a large volume of encrypted UDP on port 443 toward consumer networks in the EU. The node does not proxy third-party traffic, does not participate in file sharing, and keeps no content — only connection metadata for capacity planning (see legal & contact).
Node facts
| Hostname | hub.irdm.io |
|---|---|
| Location | Alkmaar, North Holland, NL — 52.63 N, 4.75 E |
| Network | AS60404 · 1 Gbit/s uplink, no rate limiting |
| IPv4 / IPv6 | 5.255.102.41 / not announced on this node |
| Services | TCP/443 — this website and the session API ·
UDP/443 — QUIC and media relay |
| Maintenance | Tuesdays 02:00–04:00 CET, announced 48 h in advance |
Operations
Monitoring is internal: latency, loss and session counts are sampled every 15 minutes and rolled up per region. The operations console is restricted to staff accounts and requires SSO — an unauthenticated request is rejected.
Operations console
Availability and incident history. Not exposed publicly — staff reach it over a private tunnel with single sign-on.
How the relay works
Transport choices, congestion control and the reasons behind UDP/443.
Peering & reachability
Uplink, exchange presence and measured RTT to major European networks.