Efficiently interconnecting computing nodes to enable use of high-radix network switches

Patent metadata

Field	Value
Publication number	US-11832033-B2
Application number	US-202117243534-A
Country	US
Kind code	B2
Filing date	Apr 28, 2021
Priority date	Nov 3, 2020
Publication date	Nov 28, 2023
Grant date	Nov 28, 2023

Abstract

A system for efficiently interconnecting computing nodes can include a plurality of computing nodes and a plurality of network switches coupled in parallel to the plurality of computing nodes. The system can also include a plurality of node interfaces. Each computing node among the plurality of computing nodes can include at least one node interface for each network switch among the plurality of network switches. The plurality of node interfaces corresponding to a computing node can be configured to send data to another computing node via the plurality of network switches. The system can also include a plurality of switch interfaces. Each network switch among the plurality of network switches can include at least one switch interface for each computing node among the plurality of computing nodes. A switch interface corresponding to the computing node can be coupled to a node interface corresponding to the computing node.

First claim

What is claimed is:

1. A system for efficiently interconnecting computing nodes in a datacenter, comprising:
a plurality of computing nodes including a source node and a destination node;
a plurality of network switches coupled in parallel to the plurality of computing nodes;
a plurality of node interfaces for each computing node among the plurality of computing nodes, wherein the plurality of node interfaces send and receive data communications among the plurality of computing nodes in the datacenter via the plurality of network switches, wherein the plurality of node interfaces send and receive data communications without the source node performing serialization and without the destination node performing deserialization; and
a plurality of switch interfaces, wherein each network switch among the plurality of network switches comprises a switch interface for each computing node among the plurality of computing nodes, and wherein the switch interface corresponding to a computing node is coupled to a node interface corresponding to the computing node.

2. The system of claim 1 , wherein each of the plurality of node interfaces is configured to receive a serial data stream from a data source and send the serial data stream to another computing node via a network switch among the plurality of network switches.

3. The system of claim 1 , wherein:
the plurality of network switches are optically coupled to the plurality of computing nodes;
the plurality of node interfaces comprise a plurality of node optical transceivers, each node interface among the plurality of node interfaces comprising at least one node optical transceiver; and
the plurality of switch interfaces comprise a plurality of switch optical transceivers, each switch interface among the plurality of switch interfaces comprising at least one switch optical transceiver.

4. The system of claim 3 , wherein communications between the plurality of node optical transceivers and the plurality of switch optical transceivers occur via free-space optical communications.

5. The system of claim 3 , further comprising a plurality of optical cables that optically couple the plurality of node optical transceivers and the plurality of switch optical transceivers.

6. The system of claim 1 , wherein the plurality of network switches are electrically coupled to the plurality of computing nodes.

7. A system for efficiently interconnecting computing nodes in a datacenter, comprising:
a first plurality of network switches that interconnect a first plurality of computing nodes including a first computing node and a second computing node, the first plurality of network switches being coupled in parallel to the first plurality of computing nodes;
a first plurality of switch interfaces, each network switch among the first plurality of network switches comprising a switch interface for each computing node among the first plurality of computing nodes;
a first plurality of node interfaces corresponding to the first computing node, wherein the first plurality of node interfaces are configured to send a plurality of parallel data streams to the second computing node via the first plurality of network switches, and wherein the first plurality of node interfaces send the plurality of parallel data streams to the second computing node without performing serialization; and
a second plurality of node interfaces corresponding to the second computing node, wherein the first plurality of node interfaces and the second plurality of node interfaces both comprise a node interface for each network switch among the first plurality of network switches, wherein the second plurality of node interfaces are configured to receive the plurality of parallel data streams from the first computing node via the first plurality of network switches, and wherein the second plurality of node interfaces are configured to receive the plurality of parallel data streams from the first computing node without performing deserialization.

8. The system of claim 7 , wherein each node interface among the first plurality of node interfaces is configured to receive a serial data stream from a data source and send the serial data stream to the second computing node via the first plurality of network switches.

9. The system of claim 7 , wherein each node interface among the second plurality of node interfaces is configured to receive a serial data stream from a network switch and send the serial data stream to a data sink on the second computing node.

10. The system of claim 7 , wherein:
the first plurality of network switches are optically coupled to the first plurality of computing nodes including the first computing node and the second computing node;
the first plurality of node interfaces comprise a first plurality of node optical transceivers;
the second plurality of node interfaces comprise a second plurality of node optical transceivers; and
the first plurality of switch interfaces comprise a first plurality of switch optical transceivers optically coupled to the first plurality of node optical transceivers and a second plurality of switch optical transceivers optically coupled to the second plurality of node optical transceivers.

11. The system of claim 10 , wherein the first plurality of node optical transceivers are configured to communicate with the first plurality of switch optical transceivers and the second plurality of node optical transceivers are configured to communicate with the second plurality of switch optical transceivers via free-space optical communications.

12. The system of claim 10 , further comprising:
a first plurality of optical cables that optically couple the first plurality of node optical transceivers and the first plurality of switch optical transceivers; and
a second plurality of optical cables that optically couple the second plurality of node optical transceivers and the second plurality of switch optical transceivers.

13. The system of claim 7 , wherein:
the first plurality of node interfaces are electrically coupled to a first subset of the first plurality of switch interfaces; and
the second plurality of node interfaces are electrically coupled to a second subset of the first plurality of switch interfaces.

14. The system of claim 7 , further comprising:
a second plurality of network switches coupled to a second plurality of computing nodes; and
additional network switches that enable the first plurality of computing nodes to communicate with the second plurality of computing nodes, the additional network switches being coupled to the first plurality of network switches and the second plurality of network switches.

15. A system for efficiently interconnecting computing nodes, comprising:
a first plurality of network switches that interconnect a first plurality of computing nodes including a first computing node and a second computing node, wherein the first plurality of network switches are optically coupled to the first plurality of computing nodes in a parallel configuration;
a plurality of switch optical transceivers, wherein each network switch among the first plurality of network switches comprises a switch optical transceiver for each computing node among the first plurality of computing nodes;
a first plurality of node optical transceivers corresponding to the first computing node, wherein the first plurality of node optical transceivers are configured to transmit a plurality of parallel data streams to the second computing node via the first plurality of network switches, and wherein the first plurality of node optical transceivers transmit the plurality of para

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Ballani Hitesh
• Saunders Winston Allen
• Belady Christian L
• Hsu Lisa Ru-Feng
• Costa Paolo
• Carmean Douglas M

Classifications

• H04L49/10

  characterised by the switching fabric construction · CPC title

• H04Q11/0066 Primary

  Provisions for optical burst or packet networks · CPC title

• H04Q11/0005

  Switch and router aspects · CPC title

• H04Q11/0067

  Provisions for optical access or distribution networks, e.g. Gigabit Ethernet Passive Optical Network (GE-PON), ATM-based Passive Optical Network (A-PON), PON-Ring · CPC title

• H04Q11/0071

  Provisions for the electrical-optical layer interface · CPC title