Model-Driven Telemetry

Model-Driven Telemetry Overview:

Traditional SNMP polling helped build the network monitoring tools. However, SNMP is interrupt driven and has proven to be inefficient and insecure. Model-Driven Telemetry (MDT) allows administrators to tell the device exactly what to collect and how frequently. The device publishes those stats periodically. Since it’s pre-scheduled and often hardware assisted, the MDT is much more forgiving to the device.

MDT is granular and uses efficient Google Protocol buffers that minimize the overall bandwidth needs. Many new network devices from vendors such as Arista, Cisco and Juniper already support MDT.

Model-Driven Telemetry Benefits:

  • Avoids interrupt-driven polling that is inefficient
  • More granular definition of the key performance indicators (KPI)
  • Growing vendor support

Model-Driven Telemetry with Anuta ATOM:

Anuta ATOM automatically discovers the network infrastructure. Administrators can select the devices, their specific KPIs along with the frequency of collection and retention periods.

The collected statistics can be stored in a time series database (TSDB) for forensic analysis and presented to machine learning systems for comparison with baseline behavior. Anuta ATOM can then enforce closed loop automation.