Describes how to monitor network parameters using telemetry data for proactive analysis, including prerequisites, workflow, and step-by-step configuration and verification.
The use case illustrates how, with the dial-in mode, you can use telemetry data to stream various parameters about your network. You use this data for predictive analysis where you monitor patterns, and proactively troubleshoot issues. This use case describes the tools used in the open-sourced collection stack to store and analyse telemetry data.
Watch this video to see how you configure model-driven telemetry to take advantage of data models, open source collectors, encodings and integrate into monitoring tools.
Telemetry involves the following workflow:
-
Define: You define a subscription to stream data from the router to the receiver. To define a subscription, you create a sensor-group.
-
Deploy: The receiver initiates a session with the router and establishes a subscription-based telemetry session. The router streams data to the receiver. You verify subscription deployment on the router.
-
Operate: You consume and analyse telemetry data using open-source tools, and take necessary actions based on the analysis.
Before you begin
Ensure you meet these dependancies:
-
Make sure you have L3 connectivity between the router and the receiver.
-
Enable gRPC server on the router to accept incoming connections from the receiver.
Router#configure Router(config)#grpc Router(config-grpc)#port <port-number> Router(config-grpc)#commitThe port-number ranges from 57344 to 57999. If a port number is unavailable, an error is displayed.
Starting from Cisco IOS XR Software Release 7.4.1, Cisco IOS XR, release 7.4.1 introduces enhancements to the following gRPC in-band functionalities:
Multi VRF support (other than default VRF). You must enable gRPC using the following configuration:
Router(config)#grpc Router(config)#vrf <vrf-name> Router(config-vrf)#port <port-number> Router(config-vrf)#no-tlsFor more information, see the Use gRPC Protocol to Define Network Operations with Data Models chapter in the Programmability Configuration Guide for Cisco NCS 6000 Series Routers.
Stream telemetry data out of physical sub-interface.
Stream telemetry data out of bundle interface and sub-interface.
This enhancement to in-band telemetry is supported on 20-Port 100-Gbps (2T) line cards.
Procedure
| 1. | Define a subscription to stream data from the router to the receiver. Specify the subset of the data that you want to stream from the router using sensor paths. The sensor path represents the path in the hierarchy of a YANG data model. This example uses the native data model Example: |
|
| 2. | Subscribe to telemetry data that is streamed from a router. A subscription binds the sensor-group and sets the streaming method. The streaming method can be cadence-driven or event-driven. Separating the sensor-paths into different subscriptions enhances the efficiency of the router to retrieve operational data at scale. Example:
|
|
| 3. | Verify deployment of the subscription. The receiver dials into the router to establish a dynamic session based on the subscription. After the session is established, the router streams data to the receiver to create a data lake. Verify the state of the subscription. An Example:The router streams data to the receiver using the subscription-based telemetry session and creates a data lake in the receiver. |
|
| 4. | Operate on telemetry data for in-depth analysis of the network. Start Pipeline from the shell, and enter your router credentials. The streamed telemetry data is stored in InfluxDB. Example:The streamed telemetry data is stored in InfluxDB. |
|
| 5. | Use Grafana to create a dashboard and visualize the streamed data.
In conclusion, telemetry data shows that various parameters of the network can be monitored simultaneously. This data is streamed in near real-time without affecting the performance of the network. With this data, you gain better visibility into your network. |
Once completed, the router streams telemetry data to the receiver, creating a data lake for proactive analysis. You can visualize and analyze network health and performance using open-source tools such as Pipeline, InfluxDB, and Grafana.
What to do next
Review the dashboards and analysis results in Grafana and InfluxDB to proactively monitor and troubleshoot network issues.