Teredo provides IPv6 connectivity over an IPv4 network by encapsulating IPv6 packets inside UDP datagrams sent over IPv4. This design lets hosts behind NATs reach the IPv6 Internet without a native IPv6 path, using Teredo servers to help discover the NAT type and Teredo relays to forward traffic to IPv6 destinations. It does not rely on HTTP for control and it does not tunnel IPv4 over IPv6 TCP; the traffic is IPv6 over IPv4 using UDP. So the accurate description is that Teredo tunnels IPv6 over IPv4 UDP.

Keep-alive signaling. In Teredo, NAT devices and firewalls can drop mappings if there’s no traffic for a while. A Teredo Bubble Packet is a tiny UDP packet sent periodically to refresh the NAT translation and verify the path to the Teredo relay, ensuring the IPv6 tunnel stays usable even when no actual data is being transferred. It’s not about starting a tunnel, carrying data, or reporting errors—its role is to keep the tunnel alive.

In IPv6, the mechanism used to signal packet priority and quality of service is carried by the Traffic Class field. This 8-bit field is designed to carry DSCP (Differentiated Services Code Point) and ECN (Explicit Congestion Notification) values, similar to how the IPv4 ToS field was used for prioritization. By marking packets with a Traffic Class value, routers can classify and treat them with different QoS policies as they traverse the network. The Flow Label is there to help identify packets that belong to the same traffic flow, which can assist with per-flow handling and can support QoS in some scenarios, but its primary purpose isn’t to indicate priority for individual packets. Payload Length and Hop Limit serve other roles—Payload Length specifies how much data is in the packet, and Hop Limit (the IPv6 version of TTL) prevents infinite looping. So the field that replaces the ToS concept for traffic prioritization is the Traffic Class field.

IP is a best-effort delivery service. It does not guarantee that a packet will arrive, nor that packets will arrive in the order they were sent, and duplicates can occur. The network can lose, reorder, or duplicate datagrams as they traverse routers and paths. Reliability and in-order delivery are achieved at higher layers, most notably with TCP, which uses acknowledgments, retransmissions, and sequence numbers to ensure data arrives correctly and in the right order. Routing and congestion control are not built into IP itself; those functions come from the transport layer (and, in some cases, network features that work with it). So saying that IP guarantees in-order delivery isn’t accurate—the true behavior is best-effort delivery, with reliability provided by upper-layer protocols.