IEEE International Conference on High Performance Switching and Routing
11-14 May 2020 // Newark, New Jersey, USA


The IEEE International Conference on High Performance Switching and Routing 2020 is scheduled to be held online on 11-14 May 2020.




Program at Glance



Machine Learning for Data Mining in Computer Network
Segment Routing over IPv6 (SRV6) and the Network Programming Model

Kohei Shiomoto
Tokyo City University, Japan

Stefano Salsano
University of Rome Tor Vergata, Italy

Recently machine learning and deep learning have been applied to various network management tasks. In this tutorial, we learn machine learning algorithms applied to data mining in computer network management. We learn machine learning algorithms including multi-layer perceptron (MLP), auto-encoder (AE), and generative adversarial network (GAN).

The existing supervised learning algorithms require a dataset of high quality and quantity of human-annotated data for training. To minimize the human labor-intensive and time-consuming dataset annotation task, it is thus required to find a data-efficient learning algorithm/technique to build a classifier model. We should also note that anomalies are difficult to occur in practice, so the anomaly classes are usually sparse in the dataset. As such it is extremely important for the operators to deal with an unbalanced data set where a few class has only a handful of data instances while others have a lot of data instances. We discuss challenges when we apply machine learning applied to computer network problems including training and inference, data labeling costs, feature selection, anomaly detection, and few-shot learning.

A hands-on session is prepared to understand the application of machine learning to the intrusion detection system.







Segment Routing for IPv6 (SRv6 in short) is the instantiation of the Segment Routing (SR) architecture for the IPv6 data plane. SRv6 is based on loose source routing: a list of segments (represented as IPv6 addresses) can be included in the IPv6 packet headers. According to the “SRv6 Network Programming Model”, the segments can represent both topological way-points (nodes to be crossed along the path towards the destination) and specific operations on the packet to be performed in a node. Examples of such operations are encapsulation and decapsulation, lookup into a specific routing table. More in general, arbitrarily complex behaviors can be associated with an SRv6 segment, like those that are executed in a Virtual Network Function (VNF).

The SRv6 standardization activity in IETF is progressing at a good pace. Recently, several large-scale deployments of SRv6 in operator networks have been disclosed. SRv6 implementations from different vendors are available. Linux supports SRv6 since Feb 2017.

In this tutorial, we first discuss the SRv6 architecture and the SRv6 Network Programming model. We present the main use case scenarios (SRv6 Overlays, Traffic Engineering, Fast Restoration). We introduce the ecosystem of SRv6, which includes different hardware and software implementations (proprietary and Open Source). We analyze the Linux kernel SRv6 implementation, also discussing its packet forwarding performance. Finally, we show how to build SRv6 based services using Linux on a Mininet emulation.