US 2017/0288953 A1 · AUTOMATIC SETUP OF FAILURE DETECTION SESSIONS

Abstract

For a network with host machines that are hosting virtual machines, a method for facilitating BUM (broadcast, unknown unicast, and multicast) traffic between a hardware switch (e.g., ToR switch) and the host machines is provided. The network has a set of host machines configured as a cluster of replicators for replicating BUM traffic from the hardware switch to the host machines. A set of network controllers establishes failure-detection tunnels for links between the hardware switch and the replicator cluster. The replicator cluster informs the set of controllers of a change in the membership of the replicator cluster to initiate an update to the active failure-detection sessions. The set of network controllers communicates with the replicator cluster and a ToR switch to establish bidirectional forwarding detection (BFD) sessions between one or more replicator nodes in the replicator cluster and the ToR switch.

Claims20

Dependent claims

A system comprising:

a first set of host machines, each host machine hosting a set of virtual machines;
a hardware forwarding element;
a second set of host machines for facilitating broadcast, unknown unicast, and multicast (BUM) network traffic between the hardware forwarding element and the first set of host machines;
a set of network controllers for establishing failure-detection tunnels between the hardware forwarding element and the second set of host machines, wherein the set of network controllers initiates an update to a set of active failure-detection sessions over the established failure-detection tunnels when the second set of host machines report a change to a state of the second set of host machines.

1 more element

A method for a set of network controllers to establish a failure-detection tunnel between a hardware forwarding element and a managed forwarding element on a host machine, the method comprising:

receiving state information sent from the host machine, the state information comprising at least information regarding a tunnel endpoint on the host machine;
deriving host machine information from the state information;
sending the host machine information to the hardware forwarding element, said host machine information to be used by the hardware forwarding element for (i) determining whether a failure-detection tunnel will be created and (ii) establishing a first unidirectional tunnel from the hardware forwarding element to the managed forwarding element;
receiving unidirectional tunnel information regarding the first unidirectional tunnel, the unidirectional tunnel information comprising at least information regarding the tunnel endpoint on the hardware forwarding element;
deriving tunnel endpoint information from the unidirectional tunnel information, said tunnel endpoint information to be used by the managed forwarding element to establish a second unidirectional tunnel from the managed forwarding element to the hardware forwarding element, said first and second unidirectional tunnels to be used as a failure-detection tunnel to exchange information to detect failures in the connection between the hardware forwarding element and the managed forwarding element.

2 more elements

Dependencies are inferred from the claim text.

Automatic setup of failure detection sessions

Life of this patent

Automatic setup of failure detection sessions

Life of this patent

Abstract

Claims20