Page History

Understanding inter-domain routing in our current Internet has become significantly more complex over the last decade. The volume of both BGP speakers and their announcements has increased, but moreover, deployment of RPKI Route Origin Validation leaves operators with a less transparent view on parts of their networks and equipment.

With Rotonda, the Routing team at NLnet Labs aims to provide operators and researchers a modern, modular software stack focusing on collection and analysis of real-time routing data.

The BGP Monitoring Protocol (BMP) is the principal way of getting data from routers into Rotonda. We go into the basic concepts of BMP, its pros and cons compared to other ways of collecting data, and touch upon how recent developments of the protocol can further help with routing observability.

In this talk I will share some details on our usage of BMP in the SURF network. For our SURFinternet service we have a lot of peering sessions with external networks. To get some insights in what is happening in these peering sessions, we started looking for a robust monitoring setup that is more feature-rich then just SNMP traps. Since this is very much a work in progress, I wil share some details on what tools we looked at and what we use today, some examples of events where BMP gave us a clear advantage in pinpointing an issue and to close the talk, I’ll give some sneak peeks on what we plan to do with all the data we gather.

Society's dependence on the internet has grown over the years: The internet changed from being just a network of networks to the foundation of many social and economical transactions. This change poses new challenges in the areas of transparency, responsibility, and sustainability that were never considered in the early years of the Internet. Transparency forms the foundation for internet responsibility and sustainability. Especially when multiple intermediate networks need to be traversed while transporting data, networks need to be transparent about the manner the data is processed or how much energy is consumed during the transport of the data. Only with the right information a user can make intelligent decisions about how their data will be transported.

To provide transparency on the existing internet, we earlier developed a tool named PathVis to provide users with insights on how their network traffic traverses to its destination. With PathVis, we attempted to combine selected information from existing sources into a view of how one connects to the Internet. We knew beforehand that there are a multitude of information sources of different quality levels available but combining the information from these sources into something reliable and usable is not trivial.

To improve the quality of information about independent networks, we propose an Autonomous System Information Service (ASIS): a self-hosted approach for sharing interoperability and policy information of a communication network. The ASIS gives a network (Autonomous System or AS) the autonomy to decide whether to run such a service or not and what information they share with whom. We believe the ASIS can contribute solve some of the aforementioned challenges by facilitating internet transparency.

The ultimate objective of the programmable networking research by TNO is to create a self-optimizing network service infrastructure, based on open networking and cloud technology. In this research project we make a step towards this long-term objective by investigating state-of-the-art the programmable network telemetry and its application to monitor and optimize end-to-end performance of advanced network services. Specifically, the objective is to create a system to collect and integrate telemetry data from the network, from the cloud systems hosting applications and  –  if possible – from the applications themselves.

As a specific use case for applying programmable end-to-end performance telemetry, we select eXtended Reality (XR) services. XR services are known to be highly demanding both from network (e.g., high bandwidth demand, low jitter) and processing perspective (e.g., transcoding latency).

In our talk we will discuss the architecture of the eXtended Reality system, developed in TNO SocialXR programma, instrumented in this project with various telemetry functions. We will touch upon network, cloud and application monitoring. We will present both successful developments as well as the difficulties that required workarounds or even problems that we could not solve. Finally, we will mention on how collected telemetry data is envisioned to be linked to the XR Quality of Experience.

Resource Public Key Infrastructure (RPKI) and Route Origin Validation (ROV) adoption has increased significantly over the last couple of years. However, as we have recently seen, not every network that does ROV has the same impact on where traffic goes – for example if it is surrounded by networks that do not do ROV.

With this research we want to find out whether we can find out which networks do not do ROV yet, and whether we can rank them based on the impact they would have on the global internet if they did do ROV.

We do this by announcing a valid less specific prefix from an anycast network, and a more specific invalid announcement from one location. We use RIPE Atlas and the NLNOG Ring to then perform traceroutes to two addresses: one that is inside the less specific and inside the more specific, and one that is inside the less specific and not inside the more specific. We then compare at which hop and Autonomous System (AS) the traffic deviates, and analyze which ASes are most prevalent.