Tangled: A Cooperative Anycast Testbed

Research paper wins IM 2021 Best Experience Paper Award

Connections around planet earth viewed from space at night

The research paper “Tangled: A Cooperative Anycast Testbed” has received the Best Experience Paper Award at the 2021 IFIP/IEEE International Symposium on Integrated Network Management. The paper was the product of research led by the University of Twente in collaboration with Tesorion, the University of Passo Fundo, the University of Rio Grande do Sul, and us at SIDN Labs.

Image of the IM 2021 Best Experience Paper Award, which SIDN Labs won, among others, for the paper 'Tangled: A Cooperative Anycast Testbed'

What is Tangled?

As part of a joint research project with the University of Twente, we have developed an open anycast testbed called Tangled, which is freely available to the community. Tangled is unique because it focuses on helping researchers carry out anycast experiments, without having to spend time on setting up the infrastructure. This “as-a-service” model means that researchers can focus on their anycast experiments and on understanding effects on the Internet.

Tangled has several nodes connected to commercial ISPs and, at some locations, directly to Internet Exchange Points. Table 1 shows the locations of our nine nodes.

Table of Tangled anycast nodes locations
Table 1: Tangled anycast nodes.

Because Tangled connects to multiple Autonomous Systems (ASs), it is able to use multiple traffic engineering policies. Table 2 summarises some of the available policies and the respective neighbouring ASs.

Table 2 of the active traffic management policies
Table 2: Enabled traffic engineering policies.

We have recently added support for PATH poisoning and incorporated IXP communities into our API. We are happy to add more features to our testbed, as required by the community. You can find more details in our paper.

Why another testbed?

We got involved in Tangled because of its unique focus on helping network researchers and because it complements our own Anycast2020 testbed, which enables us to dynamically instantiate anycast services across the globe to match traffic demand. Also, we believe that deploying testbeds in the wild is the most technically representative way to continually perform anycast measurements and experiment with configurations and concepts. This is important because anycast is a key building block for our DNS services for .nl, which are a vital service for the Netherlands. At SIDN Labs, we will continue to progress our studies in this field through collaborations and applied research, in fields such as anycast catchments and their performance and the management of DDoS attacks on anycast networks. Note: this was my last blog for SIDN Labs because I’ll be continuing my career at Protonmail as of June 1, 2021.