Reducing the Cost of Network Traffic Monitoring with AI (2021)

By Petar Djukic, Maryam Amiri & Wade Cherrington, Ciena Canada

This paper describes three visionary approaches to reduce the amount of network telemetry. The problem we discuss is that internet protocol (IP) network monitoring requires collection and storage of a large volume of data. This is a big issue for service providers (SPs), but the solution is usually not thought of outside of conventional approaches. With the advent of artificial intelligence (AI) approaches, especially in estimation, interpolation and imputation, new solution avenues are becoming available.

IP network monitoring often implies the use of NetFlow or Internet Protocol Flow Information Export (IPFIX). However, as we show shortly the two alone are not enough to cover all use cases. NetFlow and IPFIX can be used to turn the network into a large collection of sensors collecting information about IP traffic, which can be used to monitor network usage, identify misconfigured network elements, identify compromised network end points and detect network attacks (Santos, 2016). However, high fidelity network monitoring with technologies such as IPFIX comes with many challenges due to the amount of data that is collected by network elements, then transmitted to where it can be stored and finally processed to get the insights that the operator is looking for. IPFIX also only gives a partial view of the network state and additional technologies are needed to build a complete view of the network.

By clicking the "Download Paper" button, you are agreeing to our terms and conditions.