Publication Details
Toward migration to SDN: Generating SDN Forwarding Rules by Decision Tree
Ryšavý Ondřej, doc. Ing., Ph.D. (DIFS)
Decision tree, Software Defined Networks, Data Mining, Netflow, Flow Table, Rule
Generation.
The deployment of Software Defined Network (SDN) switches faces various
challenges, one of them is to generate rules to be preloaded in the flow table
for performance improvement. Since the flow tables in SDN are implemented by TCAM
(ternary content-addressable memory), they have a limited capacity. On the other
hand, prepopulating them with the most common rules will reduce the flow setup
delay. This paper provides a method for the identification of the most suitable
candidate rules by observing the existing traffic and automatic generation of
OpenFlow rules that can fit in SDN tables. The rules are extracted from the
decision tree build based on the observed NetFlow traffic. The experiments showed
that this method can provide a compact set of rules matching most of the network
traffic.
@inproceedings{BUT188749,
author="Sawsan {Youssef} and Ondřej {Ryšavý}",
title="Toward migration to SDN: Generating SDN Forwarding Rules by Decision Tree",
booktitle="26th Conference on Innovation in Clouds, Internet and Networks and Workshops (ICIN)",
year="2023",
pages="16--20",
publisher="IEEE Computer Society",
address="Paris",
doi="10.1109/ICIN56760.2023.10073500",
isbn="979-8-3503-9804-5",
url="https://ieeexplore.ieee.org/document/10073500"
}