Streaming network telemetry
(also streaming telemetry)
Streaming network telemetry definition
Streaming network telemetry is the process of continuously monitoring, analyzing, and collecting data from network traffic in real-time. The data provided by streaming network telemetry lets administrators better understand their network’s behavior, troubleshoot issues, and make informed decisions when it comes to network policy.
Streaming network telemetry is a significant step up when compared to traditional network monitoring methods (which typically rely on polling or periodic sampling).
How streaming network telemetry works
Streaming network telemetry gathers telemetry data (such as traffic statistics, CPU utilization, memory use, and error rate) from network devices like routers and switches in real-time.
Unlike traditional monitoring methods, streaming network telemetry does not use a pull model for data collection. In a pull model, the monitoring system periodically requests data from the devices it monitors. Instead, streaming telemetry uses a push model — devices autonomously send data to the collector at intervals set by the operator.
Telemetry data is typically encoded in a structured format (such as Google’s Protocol Buffers) for efficient data representation. This data is sent from the device to the designated central platform using lightweight protocols designed for low-latency communications, like the User Datagram Protocol (UDP) or the gRPC Remote Procedure Calls.
Finally, the central platform processes and stores the received data. Network operators are then able to pull out and analyze the relevant telemetry information to identify potential network issues.