KubeConEu 2024 Recap and Highligts

Last week I had the privilege to be present at KubeCon | CloudNativeCon EU. This year it was hosted in the city of Paris (located in France), the city of Love! In this post I want to share my story and my personal highlights. The highlights include links to awesome talks, great tools, lovely talks with vendors and other impressions of KubeCon EU 2024.

Introduction KubeCon
A short explanation, if you are not familiar with KubeCon. The Cloud Native Computing Foundation (CNCF) conference gathers adopters and technologists from leading open source and cloud native communities. They bring together the world’s top developers, end users, and vendors and run the largest open source developer conferences. This year KubeCon / CloudNativeCon Europe was this year held in France in the city of love, Paris.

CNCF hosted colocated events
KubeCon started with the different sponsor hosted and CNCF hosted colocated events like ArgoCon, CiliumCon, Observability day and Azure Day with Kubernetes. Pitty that I had to choose. Personally I was present at the Azure Day with Kubernetes, sponsered by Microsoft that was hosted a colocated event. This day had a fully packed schedule with lots of information, tips and announcements, which can be found via this link. If you want to watch the videos because you missed them or just watch them again, click here. If you have interest in the AKS Best Practices eBook, click this link.

General KubeCon highlights
Information that was shared at the KeyNotes sessions via Priyanka Sharma and Taylor Dolezal:

• Kubernetes celebrates it's 10th birtday;
• 12.000 attendees;
• A lot focus on AI and LLM;
• One new certificates that where announced: Kubernetes and Cloud Native Security Associate (KCSA);
• KubeConEu2025 will hosted next year (1-4 April 2025) hosted at the city of Londen located in England;
• KubeCon Project updates and feedback of the End User Experience.
  

KubeCon Keynote highlights Here is my list of highlights, about the KubeCon Keynotes:

• [LLM in Action] - Priyanka Sharma from CNCF talked about the new LLM Ollama and the CNCF Artificial Intelligence Whitepaper about it.
• [KubeCon in Conversation] - Jeffrey Morgan, Paige Bailey, Timothee Lacroix and Priyanka Sharma discussed how CNCF can help in solving issues for AI like monitoring, security, observability and abstracting away hardware dependencies and management.
• [Accelerating AI Workloads with GPU in Kubernetes] - Kevin Klues from NVIDIA talked about how GPU is supported in Kubernetes. He explained the GPU support in Kubernetes, GPU sharing techniques, Resource allocation (DRA) and CPU Scale out changes (Nvidea picasso – gen ai at scale and Observability).
• [Build an open Source Platform for AI/ML] - Jorge Palma from Microsoft talked about Kaito, which is an operator that automates the AI/ML inference model deployment in a Kubernetes cluster. Kaito follows the classic Kubernetes Custom Resource Definition(CRD)/controller design pattern. User manages a workspace custom resource which describes the GPU requirements and the inference specification. For more information, click here.
• [Optimizing Performance and Sustainability for AI] - Clayton Coleman , Victor Jakubiuk, Lu Qiu, Ricardo Rocha and Susan Wue talked about the way you could implement and optimize Kubernetes. They touched the topics "Make Kubernetes Amazing for accereated workloads", "GPU-free LLM inference", "Sharing and Efficient Usage of GPU Resources" and "Speed up data loading & preprocessing for AI Training". They came to the following conclusions:
  
  1. Kubernetes is beconing the standard for AI Platforms. Make accelerated workloads run better on Kubernetes.
     
  2. Make resource allocation decisions match the usage patterns.
     
  3. Speed up data loading and pre-processing by attaching CPUs to GPU clusters.
     
  4. Choose the right specialized compute for the right AI model - make it easier for your resource scientists to iterate faster.
     
• [AI Breakthroughs Revealed] - Patrick Ohly and Cathy Zhang from Intel talked about the journey through the evolving landschape of cloud native for AI and AI for cloud native. Recently a new AI whitepaper was released. Now they are focussing on security and cost efficiency.
• [Top 10 Projects] - Taylor Dolezal from Head of Ecosystem shared the top 10 CNCF Projects by Number of End User Commits. Those projects are:
  • Backstage;
  • KCL;
  • Argo;
  • Dragonfly;
  • Prometheus;
  • K8sGPT;
  • OpenFeature;
  • Kubernetes;
  • Envoy;
  • Containerd. To make commits easier, there is a link shared "Zero to Merge" to learn how to make your first contributions to CNCF Projects.
• [Revolutionizing cloud native architectures with Web assembly] - Michelle Dhanani, Kai Walter and Ralph Squillace talked about Web assembly a method that is quicker then VM's and containers. They announced SpinKube, the Hyper-efficient serverless on Kubernetes powered by WebAssembly, which is contributed to the CNCF today. Click here to learn more.
• [CNCF Community Awards Ceremony] from the End User TAB - the finalists where presented: Cern, Shopify and Expedia. The number one is Cern.
• [Building a IT Green: A Journey of Platforms, Data, and Developer Empowerement] - Gualter barbas baptista talked about the ongoing efforts to monitor and minimize the ecological impact of cloud-based applications. Central to this endeavor is collecting data on energy usage and emporing developers as agents of change to reduce the ecological costs of digitalization. He explained this at three levels:
  1. Central platform operations to maximize node usage on Kubernetes clusters (scale down at night);
  2. Providing tools for develeopers to ease workload scheduling and scaling (using Virtual Pod Autoscaler in stead of Horizontal Pod Autoscaler);
  3. Extending application monitoring for GreenOps with Kepler, Prometheus and Grafana (visibility information).
     

KubeCon CloudNative sessions highlights Here is my list of highlights, of the followed sessions at KubeCon:

• [Flux and Ecosystem] - Stefan Prodan talked about how importend it is to contribute together. Also the roadmap was explained and the future topics that will be investigated;
• [Open telemetry] - Trask Stalnaker and Austin Parker explained the project updates, next steps and AMA. Possibility to engage with OpenTelemetry contributors, ask questions about the project, and receive direct responses from the maintainers who where present at the event.
• [Future of Intelligence Cluster Ops] - Rajas Kakodkar and Amine Hilaly talked about how AI can empowr cluster operations with Kubernetes controlles backed by LLMs. It described the stages of data processing, fine-tuning LLMs and integrating them with K8s controllers and CRDs.
• [Cilium] - Liz Rice gave a talk was about connecting, observing and securing Service Mesh and Beyond with eBPF.
• [KEDA - saving Millions over the years] - together with Solene Butruille we dived into KEDA, what it is and how it works. Then, we have taken a look to the use case, using KEDA to perform event driven autoscaling down of resources. After that evaluate the metrics, learned how to evaluate cost savings in forms of a demo.
• [Dapr 2024] - Josh van Leeuwen and Yaron Schneider first talked about what Dapr. A set of developer-centric APIs for building fault-tolerant, securing and portalbe applications that run in the cloud or on the edge. Further they covered an overview of the project and show several upcoming initatives for 2024 from a maintainer perspective, including deploying Dapr in a shared model, a new API for scheduling massively scalable reliable tasks and a larger focus on application level security.
• [Kubernetes in the Confidential Computing] - Gilles Seghaier and Nayani Parameshwari talked about embarking on the captivating journey of leveraging Kubernetes in a multi-cloud setting for Secure Multiparty Computation (sMPC). They explained the withness the wondersof confidential computing, from secret sharing to collaborative computing, all empowered by Kubernetes. sMPC, a cryptographic technique for secure joint computation over private inputs, takes center stage, ensuring robust data security in a distributed, dynamic environment. Independent Kubernetes clusters materialize as secure parties, collaborating and computing seamlessly without revealing the secret input provided by a external client.
• [From RUM to Front-End Observability with OpenTelemetry] - Purvi Kanal talked about RUM Tooling is. RUM tooling focusses on high level analytics and performance metrics, wheareas front-end observability is focused on finding and fixing problems affecting users by sending telemetry that is tailored to your specific use case. As our browser applications have gotten much more complex over the last decade, our observability capabilities have not scaled at the same time pace. This is where front-end observability closes the gap. This session has given an overview how to use tracing with OpenTelemetry to start surfacing unknown-unknwons whether you're using React, Angular or something else. Together with the best practices out of the field, it is giving insights and helping in fix production issues.
• [Cloud Native Sustainability] - Kristina Devochko and Antonio Di Turi talked about the CNCF Technical Advisory community Group for Environmental Sustainability (TAG ENV), a community group dedicated to cultivating a more eco-friendly CNCF landscape. It is important that you'll recognize sustainability not just as a concept but as an urgent technical challenge demanding attention. Projects they are working on are:
  
  • Update and optimize Landschape Document v2;
  • Green scraper;
  • Kubernetes Best Practices for Sustainability;
  • Benchmarking Whitepaper;
  • Learning path;
  • User Stories. Tools like Kepler helping in getting metrics so elements can be scanned and correct being evaluated. Falco is the first CNCF Projects which has been evaluated. More will follow.
• [Open Policy Agent (OPA) intro and Deep Dive] - Anders Eknert and Xander Grzywinski talked about the introduction of OPA and showed in a deepdive sessie, which topics are currently being built. They also showed the roadmap, which can be found at this link.
• [Fortifying AI Security in Kubernetes with Confidential Containers (CoCo)] - Suraj Deshmukh and Pradipta Banerjee had an interesting talk. AI models have become valuable intellectual property that can provide organizations with a competitive edge. Users are searching for ways to secure their AI models without implicitly trusting third-party platform providers. While encryption is available to keep models secure when they're stored & transferred, they're still decrypted & loaded into memory during inferencing, potentially exposing them to unintentional or intentional exfiltration. This is where "Confidential Computing" comes in. This technology encrypts memory to protect data in use. Confidential Containers (CoCo) is an CNCF sandbox project that aims to bring confidential computing to Kubernetes. The Kubernetes AI/ML ecosystem is mature & offers many AI/ML training & inferencing options. The focus is on using CoCo with Kserve project to show how CoCo strengthens AI model protection. Apart from inferencing, they have explorerd broader applicaton of CoCo, emphasizing its role in providing general memory protection for foundational platforms.
• [Memory Armor for SPIRE: Fortifying SPIRE with Confidential Containers] - Matthew Bates and Suraj Deshmukh talked about SPIRE. This is a production-ready reference implementation of SPIFFE, a CNCF project for workload identity. In the SPIRE architecture, the SPIRE server holds considerable importance, hence securing and safeguarding it is critical. Confidential Containers (CoCo) is an emering CNCF project that promisis confidentiality for application memory. This is demonstrated in an frontend application. Learning points is about the SPIRE trust model, CoCo and iets ecosystem. It is demonstrated how it can be used to protect SPIRE infrastructure and provide even further depths of security.
• [Microsoft Azure - Future of Confidential Containers (CoCo), Microsoft Defender and eBPF] - Michael Withrow very intresting conversation in what is happening in the security field of Microsoft. More will come, stay tuned.
• [OC3 Community Edgeless Systems Event] - Special event about Confidential Compute. Presentations where shared about the latest updates from Intel, AMD and Constellation.
  
  

Wrapping up
Learned a lot of new things. Met a lot of old and new friends in person again! Did a lot of networking with people from Tigera, ASML, Thales, Codefresh, Aqua, Intel, Microsoft Azure, GitHub, Isovalent, Solo.io, Elastic, Dynatrace, OpenSearch, Diagrid, Grafana Labs, groundcover, O'Reilly, Venafi, and more. I was not able to always attend the sessions that I had planned for. Luckily I but also you can watch those sessions back when they come available online.
Keep your eyes open at the CNCF Youtube channel here and catch new talks that get published soon! If you have any questions, please contact me.