Enhancements to Hardware Timestamp

Telemetry messages carry a timestamp per interface to indicate the time when data is collected from the hardware. With this feature, the support for hardware timestamp is extended to MPLS Traffic Engineering (MPLS TE) counters, Segment Routing for Traffic Engineering (SR-TE) interface counters, protocol statistics, and bundle protocol counters.

Hardware Timestamp

Push Cached Generic Counters Data for Telemetry

This feature streams telemetry data for generic counters using the data producer to push data from the source using the telemetry push library. This push mechanism ensures that any change to the cache streams the latest data to the collector as an event-driven telemetry notification.

This feature introduces support for the following sensor path:

Cisco-IOS-XR-infra-statsd-oper:infrastatistics/

interfaces/interface/cache/generic-counters

Release 7.3.3

Target-Defined Mode for Cached Generic Counters Data

AI-driven telemetry (ADT)

This feature leverages machine learning to detect and retrieve important network-state changes on the router. Relevant data is filtered and exported to the network management system for analysis or troubleshooting purposes.

ADT significantly simplifies the configuration of streaming telemetry, and you are no longer required to manually choose sensor paths or tune the cadence at which counters have to be collected.

Build Intelligence on the Router Using AI-Driven Telemetry

Stream Digital Optical Monitoring (DOM) Data

This feature streams fiber optic transceiver parameters such as optical input or output levels, temperature, laser bias current, supply voltage, receiver power, bias threshold in real-time. This helps network operators to easily locate a fiber link failure, thereby simplifying the maintenance process, and improving overall system reliability.

Release 7.3.1

Sensor Path

Hardware timestamp

Whenever periodic statistics are streamed, the collector reads the data from its internal cache, instead of fetching the data from the hardware.

When the data is read from the cache, the rate at which data is processed shows spikes because the timestamp from the collector is off by several seconds. With hardware timestamping, the inconsistencies that are observed when reading data from the cache file is removed.

Release 7.3.1