Reliable credit-based communication over long-haul links

Patent metadata

Field	Value
Publication number	US-11929934-B2
Application number	US-202217730246-A
Country	US
Kind code	B2
Filing date	Apr 27, 2022
Priority date	Apr 27, 2022
Publication date	Mar 12, 2024
Grant date	Mar 12, 2024

Abstract

A communication apparatus includes input circuitry, an encapsulator, transmission circuitry and flow control circuitry. The input circuitry is to receive packets from a data source in accordance with a first communication protocol that employs credit-based flow control. The encapsulator is to buffer the packets in a memory buffer and to encapsulate the buffered packets in accordance with a second communication protocol. The transmission circuitry is to transmit the encapsulated packets over a communication link in accordance with the second communication protocol. The flow control circuitry is to receive from the encapsulator buffer status indications that are indicative of a fill level of the memory buffer, and to exchange credit messages with the data source, in accordance with the credit-based flow control of the first communication protocol, responsively to the buffer status indications.

First claim

The invention claimed is:

1. A communication apparatus, comprising:
input circuitry, to receive packets from a data source in accordance with a first communication protocol that employs credit-based flow control;
an encapsulator, to buffer the packets in a memory buffer and to encapsulate the buffered packets in accordance with a second communication protocol;
transmission circuitry, to transmit the encapsulated packets over a communication link in accordance with the second communication protocol; and
flow control circuitry, to receive from the encapsulator buffer status indications that are indicative of a fill level of the memory buffer, and to exchange credit messages with the data source, in accordance with the credit-based flow control of the first communication protocol, responsively to the buffer status indications.

2. The communication apparatus according to claim 1 , wherein the flow control circuitry is to exchange the credit messages irrespective of actual transmission of the packets over the communication link.

3. The communication apparatus according to claim 1 , wherein a round-trip delay over the communication link is larger than a maximal round-trip delay supported by the credit-based flow control of the first communication protocol.

4. The communication apparatus according to claim 1 , wherein the first communication protocol is InfiniBand.

5. The communication apparatus according to claim 1 , wherein the second communication protocol is Remote Direct Memory Access over Converged Ethernet (RoCE).

6. The communication apparatus according to claim 1 , wherein the second communication protocol is Transmission Control Protocol (TCP).

7. The communication apparatus according to claim 1 , wherein the packets received from the data source are layer-2 packets, and wherein the encapsulator is to encapsulate the buffered layer-2 packets onto a layer-4 transport.

8. The communication apparatus according to claim 1 , wherein the transmission circuitry is to encrypt the encapsulated packets prior to transmission over the communication link.

9. The communication apparatus according to claim 8 , wherein the transmission circuitry is to apply layer-3 encryption to the encapsulated packets.

10. The communication apparatus according to claim 1 , and comprising a management module, to receive network-topology messages via the input circuitry, and to forward the network-topology messages transparently over the communication link for processing by a peer communication apparatus.

11. The communication apparatus according to claim 1 , and comprising a management module, to receive network-topology messages via the input circuitry, and to process the network-topology messages locally so as to present the communication apparatus as a managed network device.

12. The communication apparatus according to claim 1 , wherein the packets are associated with multiple flows, and wherein the encapsulator is to preserve an order of the packets within a given flow, and to permit deviation from the order between packets associated with different flows.

13. The communication apparatus according to claim 1 , wherein the packets are associated with multiple quality-of-service (QoS) classes, and wherein the flow control circuitry is to exchange the credit messages independently for each of the QoS classes.

14. A communication method, comprising:
receiving packets from a data source in accordance with a first communication protocol that employs credit-based flow control;
buffering the packets in a memory buffer and encapsulating the buffered packets in accordance with a second communication protocol;
receiving buffer status indications that are indicative of a fill level of the memory buffer, and exchanging credit messages with the data source in accordance with the credit-based flow control of the first communication protocol, responsively to the buffer status indications; and
transmitting the encapsulated packets over a communication link in accordance with the second communication protocol, wherein a round-trip delay over the communication link is larger than a maximal round-trip delay supported by the credit-based flow control of the first communication protocol.

15. The communication method according to claim 14 , wherein exchanging the credit messages is performed irrespective of actual transmission of the packets over the communication link.

16. The communication method according to claim 14 , wherein the first communication protocol is InfiniBand.

17. The communication method according to claim 14 , wherein the second communication protocol is Remote Direct Memory Access over Converged Ethernet (RoCE).

18. The communication method according to claim 14 , wherein the second communication protocol is Transmission Control Protocol (TCP).

19. The communication method according to claim 14 , wherein the packets received from the data source are layer-2 packets, and wherein encapsulating the buffered packets comprises encapsulating the buffered layer-2 packets onto a layer-4 transport.

20. The communication method according to claim 14 , further comprising encrypting the encapsulated packets prior to transmission over the communication link.

21. The communication method according to claim 20 , wherein encrypting the encapsulated packets comprises applying layer-3 encryption to the encapsulated packets.

22. The communication method according to claim 14 , further comprising receiving network-topology messages and forwarding the network-topology messages transparently over the communication link.

23. The communication method according to claim 14 , further comprising receiving network-topology messages and processing the network-topology messages locally so as to present a managed network device.

24. The communication method according to claim 14 , wherein the packets are associated with multiple flows, and wherein transmitting the encapsulated packets comprises preserving an order of the packets within a given flow, and permitting deviation from the order between packets associated with different flows.

25. The communication method according to claim 14 , wherein the packets are associated with multiple quality-of-service (QoS) classes, and wherein exchanging the credit messages is performed independently for each of the QoS classes.

Assignees

• Mellanox Technologies Ltd

Inventors

• Liss Liran
• Bashan Ortal
• Levy Aviad
• Levi Lion

Classifications

• H04L47/39 Primary

  Credit based · CPC title

• H04L47/10

  Flow control; Congestion control · CPC title