Orchestrating Probabilistic In-band Network Telemetry for Network Monitoring

Abstract

In-band Network Telemetry (INT) is a new network measurement method that emerged with the development of Software Defined Network (SDN). Probabilistic In-band Network Telemetry (PINT) is a probabilistic version of INT, which can provide similar network visibility as INT at a smaller overhead cost. In order to obtain global network information, it is necessary to orchestrate network telemetry tasks. Current network telemetry orchestrating methods are based on INT and are not suitable for PINT. In order to use PINT to obtain global network telemetry information, we need to analyze the network telemetry orchestration problem based on the characteristics of PINT and propose approaches. In this paper, we analyze the Probabilistic In-band Network Telemetry Orchestration problem and propose a heuristic algorithm considering some key telemetry factors, which can generate high-quality solutions. In our evaluation experiments, we observe that the heuristic algorithm can produce solutions with maximum network visibility, good information freshness, and low overhead.