Ultra reliable segment routing
SR-enabled network nodes capable of replicating and recombining data packets in a manner that enables reliable, low-latency communications. In an example embodiment, a replicator node transmits to a combinator node multiple copies of a payload over different respective network paths, with the SR headers of the corresponding packets each having a replication segment identifier or a respective SID stack that includes the replication segment identifier. The combinator node delivers/forwards to the corresponding application/destination only the first-to-arrive payload copy and discards any subsequent payload copies based on the replication segment identifier. Some embodiments may beneficially reduce latency and packet loss concurrently and consistently. For example, packet loss may be reduced due to the transmission of multiple copies of the same payload over multiple network paths. Effective latency may be reduced due to the selection of the first-to-arrive payload copy for delivery/forwarding.
1. An apparatus comprising a first node that comprises packet-processing circuitry and a memory, the first node being configured to receive from a second node a packet set of two or more egress packets in response to the second node receiving an ingress packet, each of the two or more egress packets being transmitted via a different respective network path connecting the first and second nodes, wherein:
each egress packet of the two or more egress packets comprises:
a copy of a payload of the ingress packet;
a packet header comprising a replication segment identifier (RSID) identifying a replication segment, wherein the RSID is based on the payload of the ingress packet; and
the memory is configured to store each egress packet received from the second node; and
the packet-processing circuitry is configured to determine whether or not to delete the egress packet from the memory based on the egress packet's RSID, wherein:
for each ingress packet, the two or more egress packets have the same RSID; and
the length of the RSID in the egress packets is different for at least two different ingress packets.
2. The apparatus of claim 1 , wherein the packet-processing circuitry is configured to:
copy the payload of the egress packet into a corresponding outgoing packet if the packet-processing circuitry determines that the egress packet is a first-to-arrive packet of the packet set; and
discard the egress packet if the packet-processing circuitry determines that the egress packet is a subsequent packet of the packet set.
3. The apparatus of claim 2 , wherein the first node is configured to transmit the outgoing packet to a third node or deliver the outgoing packet to an application.
4. The apparatus of claim 2 , wherein the packet-processing circuitry is configured to compute a padding value for a packet header of the outgoing packet.
5. The apparatus of claim 1 , further comprising user equipment configured to implement the first node.
6. The apparatus of claim 1 , wherein the packet-processing circuitry is configured to save in a lookup table the RSID of the egress packet if the packet-processing circuitry determines that the egress packet is a first-to-arrive packet of the corresponding packet set.
7. The apparatus of claim 6 , wherein the packet-processing circuitry is configured to delete the saved RSID from the lookup table in response to an earlier of (i) a run-out of an expiration timer associated with the saved RSID and (ii) receipt of all of the two or more packets of the packet set.
8. The apparatus of claim 1 , wherein the first node further comprises a packet multiplexer to connect the memory to the different network paths.
9. The apparatus of claim 1 , wherein the packet header further comprises a segment-routing identifier identifying a segment of the respective network path for the egress packet for reaching the second node.
10. A method for an apparatus comprising a first node that comprises packet-processing circuitry and a memory, the first node receiving from a second node a packet set of two or more egress packets in response to the second node receiving an ingress packet, each of the two or more egress packets being transmitted via a different respective network path connecting the first and second nodes, wherein:
each egress packet of the two or more egress packets comprises:
a copy of a payload of the ingress packet;
a packet header comprising a replication segment identifier (RSID) identifying a replication segment, wherein the RSID is based on the payload of the ingress packet; and
the memory stores each egress packet received from the second node; and
the packet-processing circuitry determines whether or not to delete the egress packet from the memory based on the egress packet's RSID, wherein:
for each ingress packet, the two or more egress packets have the same RSID; and
the length of the RSID in the egress packets is different for at least two different ingress packets.
11. The method of claim 10 , wherein the packet-processing circuitry:
copies the payload of the egress packet into a corresponding outgoing packet if the packet-processing circuitry determines that the egress packet is a first-to-arrive packet of the packet set; and
discards the egress packet if the packet-processing circuitry determines that the egress packet is a subsequent packet of the packet set.
12. The method of claim 11 , wherein the first node transmits the outgoing packet to a third node or delivers the outgoing packet to an application.
13. The method of claim 11 , wherein the packet-processing circuitry computes a padding value for a packet header of the outgoing packet.
14. The method of claim 10 , wherein user equipment implements the first node.
15. The method of claim 10 , wherein the packet-processing circuitry saves in a lookup table the RSID of the egress packet if the packet-processing circuitry determines that the egress packet is a first-to-arrive packet of the corresponding packet set.
16. The method of claim 15 , wherein the packet-processing circuitry deletes the saved RSID from the lookup table in response to an earlier of (i) a run-out of an expiration timer associated with the saved RSID and (ii) receipt of all of the two or more packets of the packet set.
17. The method of claim 10 , wherein a packet multiplexer connects the memory to the different network paths.
18. The method of claim 10 , wherein the packet header further comprises a segment-routing identifier identifying a segment of the respective network path for the egress packet for reaching the second node.