Dynamic multipathing using programmable data plane circuits in hardware forwarding elements

Patent metadata

Field	Value
Publication number	US-11522791-B2
Application number	US-202117352258-A
Country	US
Kind code	B2
Filing date	Jun 19, 2021
Priority date	Aug 23, 2019
Publication date	Dec 6, 2022
Grant date	Dec 6, 2022

Abstract

Some embodiments provide a novel method for adjusting a path for a packet flow from a source machine to a destination machine in a network. The method of some embodiments identifies a condition at a first forwarding element along a first path traversed by the packet flow through the network. The first path traverses through a hardware, second forwarding element before the first forwarding element. In some embodiments, the second forwarding element includes a programmable data plane circuit. The method, in some embodiments, uses an application programming interface (API) of the programmable data plane circuit to provide a set of parameters to the data plane circuit that cause the data plane circuit to forego selecting the first path to forward the packets of the packet flow to the destination machine and instead to select a second path, not traversing the first forwarding element, to the destination machine.

First claim

We claim:

1. A method of configuring a hardware forwarding element to distribute packet flows between a set of two or more managed forwarding elements, the method comprising:
at a computer:
receiving metrics regarding operation of each managed forwarding element;
identifying, from the received metrics, operating conditions of the managed forwarding elements;
based on the identified operating conditions, computing a set of weight values to distribute to the hardware forwarding element; and
using an application programming interface (API) command of a programmable data plane circuit of the hardware forwarding element to provide the computed set of weight values to the hardware forwarding element to use to adjust how the hardware forwarding element distributes packet flows between the managed forwarding elements.

2. The method of claim 1 , wherein the managed forwarding elements are gateway forwarding elements of a network.

3. The method of claim 2 , wherein the network is a logical network and the managed forwarding elements are part of the logical network, while the hardware forwarding element is not part of the logical network.

4. The method of claim 2 , wherein the packet flows distributed by hardware forwarding element traverse different paths to a common destination machine in the network and the weight values allows the hardware forwarding element to select between the different paths.

5. The method of claim 1 , wherein
each managed forwarding element in at least a subset of the managed forwarding elements is a software forwarding element that executes on a host computer along with compute machines,
the compute machines use resources shared by the managed forwarding elements in the subset, and
the received metrics are associated with congestion values that are at least partially based on the use of resources by the compute machines.

6. The method of claim 1 , wherein
each managed forwarding element in at least a subset of the managed forwarding elements is a software forwarding element that executes on a host computer comprising a network interface controller (NIC),
the received metrics comprises a set of statistics related to at least one of managed forwarding element characteristics, network interface controller (NIC) characteristics, and characteristics of each host computer that executes a managed element.

7. The method of claim 6 , wherein the managed forwarding element characteristics comprise at least one of a number of virtual central processing units (vCPUs) and a number of layer 2 hops between the programmable data plane circuit and the forwarding element.

8. The method of claim 6 , wherein the NIC characteristics comprise at least one of a speed of the NIC and a utilization of the NIC.

9. The method of claim 6 , wherein the host computer characteristics comprise at least one of a number of central processing unit (CPU) cores on the host computer, a speed of the CPUs of the host computer, a utilization of the CPUs of the host computer, and a memory utilization.

10. The method of claim 1 , wherein the received metrics comprise data indicating congestion at at least one particular managed forwarding element and the computed set of weight values causes the hardware forwarding element to distribute fewer packet flows through the particular managed forwarding element in order to reduce the congestion at the particular managed forwarding element.

11. A non-transitory machine readable medium for execution by a set of processing units of a computer, the program for configuring a hardware forwarding element to distribute packet flows between a set of two or more managed forwarding elements, the program comprising a set of instructions for:
receiving metrics regarding operation of each managed forwarding element;
identifying, from the received metrics, operating conditions of the managed forwarding elements;
based on the identified operating conditions, computing a set of weight values to distribute to the hardware forwarding element; and
using an application programming interface (API) command of a programmable data plane circuit of the hardware forwarding element to provide the computed set of weight values to the hardware forwarding element to use to adjust how the hardware forwarding element distributes packet flows between the managed forwarding elements.

12. The non-transitory machine readable medium of claim 11 , wherein the managed forwarding elements are gateway forwarding elements of a network.

13. The non-transitory machine readable medium of claim 12 , wherein the network is a logical network and the managed forwarding elements are part of the logical network, while the hardware forwarding element is not part of the logical network.

14. The non-transitory machine readable medium of claim 12 , wherein the packet flows distributed by hardware forwarding element traverse different paths to a common destination machine in the network and the weight values allows the hardware forwarding element to select between the different paths.

15. The non-transitory machine readable medium of claim 14 , wherein the computation of weight values based on the collected metrics and the distribution of the weight values allow the hardware forwarding element to perform dynamic weighted path selection.

16. The non-transitory machine readable medium of claim 14 , wherein
each managed forwarding element in at least a subset of the managed forwarding elements is a software forwarding element that executes on a host computer along with compute machines,
the compute machines use resources shared by the managed forwarding elements in the subset, and
the received metrics are associated with congestion values that are at least partially based on the use of resources by the compute machines.

17. The non-transitory machine readable medium of claim 11 , wherein
each managed forwarding element in at least a subset of the managed forwarding elements is a software forwarding element that executes on a host computer comprising a network interface controller (NIC),
the received metrics comprises a set of statistics related to at least one of managed forwarding element characteristics, network interface controller (NIC) characteristics, and characteristics of each host computer that executes a managed element.

18. The non-transitory machine readable medium of claim 17 , wherein the managed forwarding element characteristics comprise at least one of a number of virtual central processing units (vCPUs) and a number of layer 2 hops between the programmable data plane circuit and the forwarding element.

19. The non-transitory machine readable medium of claim 17 , wherein
the NIC characteristics comprise at least one of a speed of the NIC and a utilization of the NIC; and
the host computer characteristics comprise at least one of a number of central processing unit (CPU) cores on the host computer, a speed of the CPUs of the host computer, a utilization of the CPUs of the host computer, and a memory utilization.

20. The non-transitory machine readable medium of claim 11 , wherein weight values configure the hardware forwarding element to use two different ports that connect to at least two different paths to a common destination machine.

Assignees

• Vmware Inc

Inventors

• Kommula Raja
• Polychronopoulos Constantine
• Sridhar Thayumanavan
• Bordeleau Marc-Andre
• Choh Edward
• Gupta Ojas
• Kidd Robert
• Oikonomou Georgios
• Tidemann Jeremy

Classifications

• H04L45/14

  Routing performance; Theoretical aspects · CPC title

• H04L45/34

  Source routing · CPC title

• H04L45/24 Primary

  Multipath · CPC title

• H04L45/22 Primary

  Alternate routing · CPC title