Throughput resilience during link failover

Patent metadata

Field	Value
Publication number	US-10097462-B2
Application number	US-201615242486-A
Country	US
Kind code	B2
Filing date	Aug 20, 2016
Priority date	Apr 2, 2016
Publication date	Oct 9, 2018
Grant date	Oct 9, 2018

Abstract

Techniques disclosed herein provide an approach for providing throughput resilience during link failover when links are aggregated in a link aggregation group (LAG). In one embodiment, failure of a link in the LAG may be detected, and a Transmission Control Protocol/Interact Protocol (TCP/IP) stack notified to ignore packet losses and not perform network congestion avoidance procedure(s) for one round-trip timeout (RTO) period. In a virtualized system in particular, a virtual switch may be configured to generate events in response to detected link failures and notify TCP/IP stacks of a hypervisor and/or virtual machines (VMs) of the link failures. In turn, the notified TCP/IP stacks of the hypervisor and/or VMs may ignore packet losses and not perform network congestion avoidance procedure(s) for one RTO period.

First claim

What is claimed is:

1. A method for providing throughput resilience during link failover, comprising:
determining by a protocol layer originating at least one packet flow on a link in an aggregation of links that the link has failed, wherein the at least one packet flow on the failed link is transferred to one or more other links in the aggregation of links that are active; and
disabling one or more network congestion avoidance procedures that reduce throughput of the at least one packet flow by the protocol layer on the at least one packet flow for a period of time by causing the one or more network congestion avoidance procedures to ignore packet losses of the at least one packet flow during the period of time.

2. The method of claim 1 , wherein the period of time is one round-trip timeout (RTO) period.

3. The method of claim 2 , wherein the one or more network congestion avoidance procedures are disabled for the one RTO period beginning from when the link failure is determined.

4. The method of claim 2 , wherein the period of time is user-configurable.

5. The method of claim 1 , wherein the disabled one or more network congestion avoidance procedures include a congestion avoidance mode in which congestion control windows (cwnd) maintained for flows associated with the failed link are reduced in size.

6. The method of claim 1 , wherein links in the aggregation of links are connected to a virtual switch, and wherein the protocol layer is in a hypervisor or in a guest operating system (OS) running in a virtual machine (VM).

7. The method of claim 6 , wherein determining the link in the aggregation of links has failed includes:
receiving, by the protocol layer, from a virtual network interface controller (VNIC), a first notification of the failed link, wherein the VNIC transmits the first notification in response to receiving a second notification from the virtual switch of the failed link, and wherein the virtual switch transmits the second notification in response to receiving a third notification from the hypervisor of the failed link.

8. The method of claim 1 , wherein the determining that the link in the aggregation of links has failed includes receiving, from a network interface controller (NIC) driver, a notification of the failure.

9. The method of claim 1 , wherein, after the period of time, one or more TCP/IP network congestion avoidance procedures are performed.

10. A non-transitory computer-readable storage medium containing a program which, when executed by one or more processors, performs operations for providing throughput resilience during link failover, the operations comprising:
determining by a protocol layer originating at least one packet flow on a link in an aggregation of links that the link has failed, wherein the at least one packet flow on the failed link is transferred to one or more other links in the aggregation of links that are active; and
disabling one or more network congestion avoidance procedures that reduce throughput of the at least one packet flow by the protocol layer on the at least one packet flow for a period of time by causing the one or network congestion avoidance procedures to ignore packet losses of the at least one packet flow during the period of time.

11. The non-transitory computer-readable storage medium of claim 10 , wherein the period of time is one round-trip timeout (RTO) period.

12. The non-transitory computer-readable storage medium of claim 11 , wherein the one or more network congestion avoidance procedures are disabled for the one RTO period beginning from when the link failure is determined.

13. The non-transitory computer-readable storage medium of claim 11 , wherein the period of time is user-configurable.

14. The non-transitory computer-readable storage medium of claim 10 , wherein the disabled one or more network congestion avoidance procedures include a congestion avoidance mode in which congestion control windows (cwnd) maintained for flows associated with the failed link are reduced in size.

15. The non-transitory computer-readable storage medium of claim 10 , wherein links in the aggregation of links are connected to a virtual switch, and wherein the protocol layer is in a hypervisor or in a guest operating system (OS) running in a virtual machine (VM).

16. The non-transitory computer-readable storage medium of claim 15 , wherein determining the link in the aggregation of links has failed includes:
receiving, by the protocol layer, from a virtual network interface controller (VNIC), a first notification of the failed link, wherein the VNIC transmits the first notification in response to receiving a second notification from the virtual switch of the failed link, and wherein the virtual switch transmits the second notification in response to receiving a third notification from the hypervisor of the failed link.

17. The non-transitory computer-readable storage medium of claim 10 , wherein the determining that the link in the aggregation of links has failed includes receiving, from a network interface controller (NIC) driver, a notification of the failure.

18. The non-transitory computer-readable storage medium of claim 10 , wherein, after the period of time, one or more TCP/IP network congestion avoidance procedures are performed.

19. A system, comprising:
a processor;
a plurality of network interface controllers (NICs), wherein the NICs are grouped together in an aggregation of links; and
a memory, wherein the memory includes a program executable in the processor to perform operations for providing throughput resilience during link failover, the operations comprising:
determining by a protocol layer originating at least one packet flow on a link in the aggregation of links that the link has failed, wherein the at least one packet flow on the failed link is transferred to one or more other links in the aggregation of links that are active; and
disabling one or more network congestion avoidance procedures that reduce throughput of the at least one packet flow by the protocol layer on the at least one packet flow for a period of time by causing the one or more network congestion avoidance procedures to ignore packet losses of the at least one packet flow during the period of time.

20. The system of claim 19 , wherein the period of time is one round-trip timeout (RTO) period beginning from when the link failure is determined.

Assignees

• Nicira Inc
• Niciria Inc

Inventors

• Sreeramoju Akshay Kumar

Classifications

• H04L45/22

  Alternate routing · CPC title

• H04L47/12 Primary

  Avoiding congestion; Recovering from congestion · CPC title

• H04L41/0654

  using network fault recovery (ring fault isolation or reconfiguration in loop networks without recovery actions by a network management system H04L12/437) · CPC title

• H04L41/40

  using virtualisation of network functions or resources, e.g. SDN or NFV entities · CPC title

• Y02D30/50

  in wire-line communication networks, e.g. low power modes or reduced link rate · CPC title