Service-based node-centric ECMP health

Patent metadata

Field	Value
Publication number	US-11025534-B2
Application number	US-201916601944-A
Country	US
Kind code	B2
Filing date	Oct 15, 2019
Priority date	Oct 15, 2019
Publication date	Jun 1, 2021
Grant date	Jun 1, 2021

Abstract

Techniques are presented herein for determining equal cost multi-path (ECMP) performance health in a network. In accordance with an example embodiment, a method is provided that involves obtaining information indicating ECMP paths in a network, the network comprises of a plurality of nodes that includes at least one branch node. The method includes configuring the branch node to obtain ECMP performance metrics, and obtaining the ECMP performance metrics from the branch node. The method further includes analyzing the ECMP performance metrics to determine ECMP performance health of the branch node.

First claim

What is claimed is:

1. A method comprising at a computer device:
based on network configuration and topology information, identifying one or more branch nodes with equal cost multi-path (ECMP) paths in a network, each ECMP path comprising a plurality of nodes that includes a branch node of the one or more branch nodes, and for each branch node identified by the computer device:
configuring the branch node to obtain ECMP performance metrics associated with the ECMP path for a particular service among multiple services provided by the network, the configuring including providing, to the branch node, intent-based monitoring objects to configure the branch node to:
record the ECMP performance metrics; and
report the ECMP performance metrics in telemetry with a service tag that identifies the particular service;

based on the configuring, receiving, from the branch node, the ECMP performance metrics with the service tag and including:
a node level performance metric for the branch node;
multiple link level performance metrics for respective links associated with the branch node; and
multiple flow-level metrics with flow information for respective interfaces of the branch node; and

analyzing the ECMP performance metrics to determine ECMP performance health of the branch node as healthy, healthy with some failures, or unhealthy based on a comparison of the ECMP performance metrics with respective thresholds and information associated with flow sampling data.

2. The method of claim 1 , wherein receiving the ECMP performance metrics includes obtaining local node metrics which include the node level performance metric and obtaining monitoring metrics, wherein analyzing is performed based on the local node metrics to determine a measure of forwarding performance and the analyzing is performed based on the local node metrics and the monitoring metrics to determine a measure of forwarding performance and a measure of load distribution performance.

3. The method of claim 2 , wherein the local node metrics include, for each egress interface of the branch node, one or more of: interface status, interface errors, interface load, and interface stability.

4. The method of claim 3 , wherein the local node metrics include, for a prefix or label with ECMP egress interfaces on the branch node, one or more of: uptime of the prefix or label, time since last change, maximum paths configured, maximum paths available, consistency of a routing information base with ECMP egress interfaces, forwarding information base consistency with ECMP egress interfaces per the routing information base, label forwarding information base consistency with ECMP egress interfaces, hardware memory programming consistent with routing information base and/or forwarding information base, data plane monitoring enabled for the prefix or label.

5. The method of claim 3 , wherein the local node metrics include, for a prefix or label with ECMP egress interfaces on the branch node: the multiple flow-level metrics collected for a period of time.

6. The method of claim 3 , wherein the local node metrics include, for each ECMP egress interface, one or more of: type including physical or virtual, uptime, time since last change, counter statistics, maximum transmission unit, number of packets switched for a prefix or a label, whether bidirectional forwarding detection is enabled, whether an egress policy is enabled.

7. The method of claim 3 , wherein the local node metrics include, for each virtual interface configured on the branch node, one or more of: type of virtual interface, list of all links for each virtual interface, uptime of each link for each virtual interface, statistics from each link, number of packets switched, whether micro bidirectional forwarding detection is enabled.

8. The method of claim 2 , wherein analyzing comprises evaluating one or more of: metrics associated with egress interfaces on the branch node, error statistics, load distribution among links/paths, whether the branch node has a single point of failure.

9. The method of claim 2 , wherein:
the obtaining the monitoring metrics includes obtaining data plane monitoring results from probe packets sent by the branch node; and
the analyzing includes analyzing the data plane monitoring results.

10. The method of claim 9 , wherein the data plane monitoring results indicate whether there is one or more of: an incorrect readable label depth on the branch node, incorrect hash algorithm employed by the branch node or incorrect execution of a hash algorithm by the branch node.

11. The method of claim 2 , wherein the ECMP performance metrics further include an aggregate node level ECMP performance metric based on a plurality of node level metrics for a given prefix or label, an aggregate link level ECMP performance metric based on a plurality of link level metrics, an aggregate forwarding ECMP performance metric based on the monitoring metrics, and the flow sampling data, and wherein analyzing includes computing a measure of ECMP performance health using an expression that combines comparison results of the aggregate node level ECMP performance metric, the aggregate link level ECMP performance metric, and the aggregate forwarding ECMP performance metric with respective thresholds, and with knowledge obtained from the flow sampling data.

12. The method of claim 1 , wherein configuring comprises configuring one or more transit nodes in the network to compute a list of links that are part of ECMP in the network for monitoring ECMP performance health of Layer 2 and Layer 3 operations in the network.

13. An apparatus comprising
a communication interface configured to enable communication with one or more of a plurality of nodes in a network; and
a processor coupled to the communication interface, wherein the processor is configured to perform operations including:
based on network configuration and topology information, identifying one or more branch nodes with equal cost multi-path (ECMP) paths in the network, each ECMP path comprising a plurality of nodes that includes a branch node of the one or more branch nodes, and for each branch node identified:
configuring the branch node to obtain ECMP performance metrics associated with the ECMP path for a particular service among multiple services provided by the network, the configuring including providing, to the branch node, intent-based monitoring objects to configure the branch node to:
record the ECMP performance metrics; and
report the ECMP performance metrics in telemetry with a service tag that identifies the particular service;

based on the configuring, receiving, from the branch node, the ECMP performance metrics with the service tag and including:
a node level performance metric for the branch node;
multiple link level performance metrics for respective links associated with the branch node; and
multiple flow-level metrics with flow information for respective interfaces of the branch node; and

analyzing the ECMP performance metrics to determine ECMP performance health of the branch node as healthy, healthy with some failures, or unhealthy based on a comparison of the ECMP performance metrics with respective thresholds and information associated with flow sampling data.

14. The apparatus of claim 13 , wherein the processor is configured to perform the receiving the ECMP performance metrics by obtaining local node metrics which include the node level performance metric and obtaining monitoring metrics, and to perform the analyzing based on the local node metrics to determine a measure of forwarding performance and to perform the analyzing based on the local nod

Assignees

• Cisco Tech Inc

Inventors

• Nainar Nagendra Kumar
• Pignataro Carlos M
• Claise Benoit

Classifications

• H04L45/48

  Routing tree calculation · CPC title

• H04L47/125

  by balancing the load, e.g. traffic engineering · CPC title

• H04L45/70

  Routing based on monitoring results · CPC title

• H04L45/245 Primary

  Link aggregation, e.g. trunking · CPC title

• H04L45/24 Primary

  Multipath · CPC title