Chaos Computer Club - recent events feed

As a Junior Engineer, the consumability of a network automation tool can really make or break your experience. Some incredibly smart engineers and developers build amazing tools for specific use cases, but they can be tough to use unless you know all their quirks. This is especially frustrating when you're just starting out. That is why I want to share a small project I worked on including all the applications I used and what I learned from it. Fancy tools are great, but at the end of the day, someone has to actually use them—both now and in the future, right? Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/VA9SFE/

That's a wrap on Day1, time for the social! Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/MVDVUL/

Past approaches for maximizing the data capacity per fiber pair went for running more and more DWDM channels with grid spacings as small as possible. This meant that grid spacings shrank from 200GHz to 100Ghz and then 50GHz with some applications even going for 25Ghz. In recent years the bandwidth per channel kept increasing, as complex modulation schemes came into favor over ON-OFF-Keying which has been a staple in fiber optic communication for decades. Those increased per channel bandwidths of 400Gbps, 800Gbps and now pushing into the 1.6Tbps realm demand for lager grids to accommodate the spectrum necessary to operate such “Superchannels”. Especially the fact of coherent detection being “blind” to anything but its own wavelength has enabled interesting topologies that can omit filters altogether. Of course that comes at a cost of reduced flexibility. Learn how you can harvest some of those benefits by moving to 400GHz filters while keeping a lot of your flexibility. Keep your 10Gbps and below legacy signals on the same fiber as your fast stuff and easily reassign your bandwidth without the cost ROADMs. Made by FLEXOPTIX Research - Gert and Thomas. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/8N7Q3A/

At Hetzner we’ve historically used an Open vSwitch based data plane for connecting hundred thousands of cloud servers to the network. This has served us well for many years and mostly still does. We have however reached some limitations and wanted to improve scalability, resiliency and flexibility with a more specialized data plane that's tailored to our needs while being easy to operate and building a strong foundation for new features. When checking our options back in 2022, the team reached the conclusion that the best path to achieve this goal is to build and maintain our own highly specialized networking stack based on eBPF/XDP, and so we went on a journey to make it reality. Today, roughly three years later, we’ve implemented a versatile network stack, called `hcnet`, which handles public and private cloud networking (using VXLAN encapsulation), stateful firewalls, and provides DHCP services as well as traffic capture tooling - all of this using XDP with a control plane written in Go. To make operation’s life easier, the stack is collecting and exposing meaningful metrics and is designed to self-heal whenever possible. We’ve been using `hcnet` in our internal cloud for two years now, with every new feature getting its first real-world tests there on a daily basis, including customer-facing applications. We are looking forward to a public beta, once we have full feature parity with our existing stack. In this talk we want to provide an overview of how we’ve built the new network stack, what challenges we’ve faced and where we're hitting current limitations of XDP. As of today the most pressing challenges are support for offloading and driver maturity in general. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/VNLHD7/

For years, InfiniBand has dominated high-performance interconnects. Yet for strategic and technical reasons, there has been a huge industry ask to develop a new next-generation protocol for use in AI and HPC environments. This session introduces Ultra Ethernet, an innovative standard designed to tackle the limitations seen in RDMA around scalability, latency and congestion management, crucial for democratizing future AI/HPC build outs. Join us for a comprehensive exploration of the technologies poised to redefine Storage or GPU interconnectivity and challenge the status quo. We also discuss the future of the UEC and how collaboration among standards organizations will benefit the industry. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/W8W77A/

At DE-CIX, we're redefining network orchestration from first principles. Confronted with the limitations of existing systems, we set out to design a modern orchestration stack—one capable of managing a truly global network, seamlessly interfacing with a variety of upstream and downstream systems, supporting fully managed network devices, and, above all, delivering uncompromising reliability. In this talk, we share our transformation journey—from our starting point to our vision for the future. We'll delve into the architectural choices, design principles, and decision-making frameworks that guided our approach. We’ll also explore how we rigorously test our systems and highlight the critical lessons we’ve learned—insights that are applicable to any complex network automation project. Our solution is built on three key pillars: a tightly integrated central source of truth for configuration management, an event-driven architecture for service orchestration, and a highly scalable design that accommodates rapid global growth. To ensure stability and correctness, we've implemented a digital twin—internally dubbed Shadow—which mirrors the production environment. This shadow stack enables us to validate and test new software iterations in parallel, minimizing risk and accelerating confidence before deployment to real hardware. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/JECPAH/

As it's common practice, each new hacker event needs something weird and funky. For this years summer camp in the Netherlands called WHY2025, we build a 400G link over a single fibre. Since 400G BiDi does not exist (yet), we had to get creative and built something, which might be considered out of spec. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/KLXJS8/

As Bird is evolving over the years and we now have Bird version 3 available at our fingertips, maybe we should know the capabilities of it and what performance we can expect from it. Get some numbers in comparison with other routing stacks. Let's take a quick look on the shiny new feathers Bird prepared for us. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/XRVTP3/

How we route-leaked everything to everyone due to a fun Arista bug. A post-mortem kind of story about us changing Arista RCF function names, resulting in a global route leak. The talk includes the history of why we intended to make a change, how we rollout configurations, why it resulted in an unexpected behavior and how it was fixed. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/KQ9JWA/

This talk explores the adoption of `security.txt`, as defined in [RFC 9116](https://www.rfc-editor.org/rfc/rfc9116), that enables websites to publish security contact information in a consistent and accessible way. We begin with a brief introduction to the RFC and the motivation behind standardized vulnerability disclosure. But is this even important to the network industry? To find out, we conclude with a focused analysis of security.txt adoption among organizations represented by this conference’s attendees, highlighting real-world trends, blind spots, and where we go from here. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/WKBJGE/

This talk presents how to to offload networking tasks onto dedicated hardware/smartNICs/NPUs (network processing units) - chips, and why this is a great idea. First, we take a look at chips in general - how they are made, integrated into networks, and why time is playing to the advantage of dedicated chips vs. CPUs/software. Then, focus is on how offloading network tasks is a particularly beautiful example of the advantages of hardware-implementing logic, with examples like TCP offloading, cybersecurity applications, or traffic shaping. To be fair and balanced, there will also be a look at the downsides of chips, but of course also ideas for mitigating them. Finally, there will be a look at what's out there, and how to integrate custom chips into your systems. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/VEGBSU/

In high-performance optical communication systems transceiver health and efficiency are critical to network reliability and energy consumption. This presentation explores the powerful capabilities of Versatile Diagnostics Monitoring (VDM) features found in modern optical transceivers (beyond the speed of 100G) with a particular focus on the Thermoelectric Cooler (TEC) current metrics. By analysing TEC current alongside temperature and cable length data we should be able to identify the optimal operating conditions that will minimise power consumption while maintaining performance. In order to prove this thesis we developed a couple of software prototypes to perform the data analysis. Using real-world VDM data this talk will demonstrate how to evaluate and visualise transceiver efficiency in terms of Watts, uncovering practical insights for engineers aiming to design or operate greener, more efficient optical networks and finally save energy. Made by FLEXOPTIX Research - Gerhard and Thomas. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/NRBPX9/

After discussing building a 1 Tbps/1 Gpps load generator on commodity x86-64 hardware, one of the most common follow-up topics was the reasoning behind some of the decisions. And here I've realised that it is not commonly understood that modern systems are more complex than classic multi-socket NUMA systems, and that if you don't consider some of their peculiarities, you won't achieve the expected performance. In this talk, I want to explore what happens when you break those assumptions and how you can identify them. All examples would be based on the assumption that you are using Cisco T-Rex for load generation and fd.io VPP as a target load. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/VJJ9RC/

Last year we migrated our datacentre networks from a flat layer 2-based architecture to EVPN-VXLAN. As an organisation whose primary technical background is in software, a network project of this complexity has been a journey into new and uncharted territory. Our previous network design had been showing signs of reaching its limits for several years, so in late 2023 we started designing a replacement based on EVPN-VXLAN, terminated directly on our Linux hypervisors. After not quite nine months of development the project culminated in an intensive week-long migration in our production datacentre – with zero downtime for our customers. In this talk I'll explore our experiences developing and operating an EVPN-based network in production coming from a software background, including debugging tales from the lab and some of our more peculiar technical decisions, as well as the unexpected surprises along the way and ongoing pain points we're still dealing with a year down the line. What *does* happen when software people try to use EVPN anyway? Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/BWHD38/

Do you have probes installed to monitor your network border? Ever struggled to estimate the fallout-perimeter of a probe alerting quality-deterioration? We show an example how to firmly map probes to interconnect points and secondly, how to identify which other traffic is likely affected by the same problem. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/YGMU8X/

Automation is at the heart of Inter.link's operations. Our team manages complex software that automates the entire customer journey, from correctly accepting orders for services such as IP Transit or DDoS and translating them into precise network configurations, to setting up billing and robust monitoring. This automation is critical, allowing us to rapidly develop and deploy new features while maintaining operational excellence. Achieving this level of automation required investment in our software environment and tooling. Over time, we have significantly evolved our tooling, emphasizing easy and consistent environment setups, and we have implemented robust CI/CD pipelines that ensure changes are delivered to our customers safely, consistently, and quickly. In this talk, we will provide a behind-the-scenes look at Inter.link's software stack. We will then explore challenges we faced that impeded quick, consistent and safe deployments, and how they informed and evolved our software environment. We will show the solutions we adopted, and how they became principles that provide guidance for the team. We will also finish off with a glimpse into some of our future initiatives. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/RZJPC9/

This talk would be held by my two colleagues: Michael Bayr (artcodix) and Gerhard Bader (Yorizon Cloud) Yorizon Cloud is the joint venture of HOCHTIEF PPP Solutions and Thomas-Krenn.AG In these times of global political and technical threats and disruptions, we are driven by our spirit of digital sovereignty when we design a new European, de-centralized, sustainable and custom-made cloud infrastructure. In their talk, my two colleagues would give an insight into the technologies we apply, open source based and European to the core. One basic focus will be the network technology and the standardized stack that we use. We will be happy to present this talk at DENOG17, feel free to contact me personally if you would like to get in touch with our two potential speakers. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/F9DVEL/

Network operations engineers maintain the backbone of our digital experiences, ensuring robust connectivity and managing complex routing. Yet, even with great network health, where are the current limits of today’s routing technology? What does it mean when your manager calls you and says “Our customers can’t watch Netflix!”? Can we measure how the end users experience the high variety of OTT-services, beyond classical speedtests? In this talk, we will investigate real-world scenarios where network issues, even seemingly minor ones, impacted video QoE. We'll demonstrate how solutions to these problems often go miles beyond classic NOC-room resolutions, requiring a deeper understanding of streaming protocols, user-centric performance, and QoE. Drawing on insights from standardized QoE metrics, we'll demonstrate how understanding the user's perception of video quality provides a critical new layer of visibility. This talk aims to equip network engineers with insights to proactively diagnose and address the hidden causes of poor customer video experience. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/WJ8JGT/

EVPN is (now) a well know Service Framework and is widely used as a replacement for VPLS ELAN Services and even some IP-VPN Services. EVPN was able to also do P2P ELINE Style services for quite some time, but with the finalization of RFC9744 it got some very interesting capabilities that can especially help aggregation networks to reduce scaling problems. The talk will give a very brief primer on how the EVPN Service Framework is being setup to allow for easy extension, focuses on the Flex Part of the new extension and will give some examples on how this impacts some (near to) real world scenarios. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/FNVVFQ/

Live video stresses every layer of a CDN: traffic spikes hit terabit scale in seconds, latency budgets shrink to single-digit seconds, and a single congested peering link can stall millions of viewers. This talk walks through how we built a custom Golang edge stack that now serves 5.5 Tbps across five German PoPs. We’ll unpack the caching hierarchy, peer-aware user steering that exploits BGP and ECMP, and the trade-offs between channel affinity, cache size, and uplink utilization. We'll also touch on observability with Prometheus and client-app telemetry enriched by CMCD. Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/SZZGTP/

Welcome to DENOG17! Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/YPAXF8/

Welcome to your first DENOG conference. In this session, we will guide you through the community, the event and everything DENOG! Licensed to the public under http://creativecommons.org/licenses/by/4.0 about this event: https://pretalx.com/denog17/talk/KFLJDB/

Closing talk for eth0:2025 about this event: https://c3voc.de

Jhaand shows us how he designed and built House 2, a complex art project involving electronics and microcontrollers. about this event: https://c3voc.de

Licensed to the public under https://creativecommons.org/licenses/by/4.0/ about this event: https://c3voc.de

Licensed to the public under https://creativecommons.org/licenses/by/4.0/ about this event: https://c3voc.de