Kubernetes à la Parisienne: AWS at KubeCon EU 2024 recap

Au Revoir, Paris! We truly enjoyed our time with the open source community at KubeCon + CloudNativeCon Europe 2024. AWS conference speakers talked about Karpenter, Argo CD apps, multi-tenant scalable Prometheus with Cortex, eBPF, chaos engineering, Kubernetes controllers, OpenTelemetry, and zonal outages. We are passionate about the latest open source innovations that make it easy for you to build and run open source software in the cloud. ☁️🌐

If you missed one of our presentations because you were caught up in conversation 🗣️, picking up some swag in the Solutions Showcase 🛍️, or taking a much-needed break ☕, we've got you covered. Our presentations are available online and listed below. 👇

AWS presenters at ArgoCon

Harnessing Karpenter: Transforming Kubernetes Clusters with Argo Workflows - Carlos Santana & Raj Saha, AWS

Discover the future of Kubernetes cluster scaling with Karpenter, the latest and most rapid Kubernetes-native cluster autoscaler, now part of the CNCF ecosystem. While Cluster Autoscaler remains widely used among Kubernetes enthusiasts, this session introduces a groundbreaking approach to transitioning your worker nodes and pods to Karpenter with ease and efficiency. Join us for an interactive demonstration where we'll explore the powerful synergy of Argo Workflows and Karpenter. You'll learn how to seamlessly migrate your Kubernetes resources, capitalizing on Argo Workflows' flexibility and its unique capability to execute CI pipelines within the cluster. This not only enhances your security posture but also adeptly manages challenges such as intermediate failures and time-intensive tasks in large-scale node roll-overs. Click here to watch the replay!

Adobe/AWS: Key Takeaways from Scaling Adobe's CI/CD Solution to Support >50K Argo CD Apps - Andrew Lee, AWS & Vikram Sethi, Adobe

Adobe Flex is a CI/CD solution started inside Adobe in 2022 with widespread adoption in Q2 2023. The broader adoption of the Flex solution quickly revealed some stability and scalability challenges with Flex struggling at around 1500 Argo CD applications. This, along with other scalability challenges faced by other AWS customers, led to a joint community effort between AWS, Adobe, Akuity, and other community members to also kick off the #argo-sig-scalability working group to collaborate on common patterns and practices around using Argo at scale. Fast forward to today, Adobe has been able to successfully scale Argo CD to run more than 9000 Argo CD applications comfortably and can easily support 5X more by adding scale on demand. In this talk, we will talk about how Adobe partnered with AWS to solve the stability and scalability challenges in getting to current scale and design a linearly scalable multi-tenant sharding architecture to easily add 5X more scale on demand, and beyond. Click here to watch the replay

AWS presenters at KubeCon

Cortex Intro: Multi-Tenant Scalable Prometheus - Ben Ye

Cortex provides horizontally scalable, highly available, multi-tenant, long term storage for Prometheus. In this talk, Ben will do an introduction of Cortex architecture and project status. He will also walk through those new features added to Cortex throughout 2023 and how to utilize them efficiently in production. Click here to watch the replay!

Troubleshooting Hidden Performance and Costs in Network Traffic Across Multiple AZs with eBPF - Shahar Azulay, Groundcover & Nirmal Mehta

Spanning Kubernetes Clusters across multiple Availability Zones is common when optimizing for resiliency but introduces challenges like network performance and costs when workloads communicate with each other across AZs. AZs are designed for low roundtrip latency between different AZs in the same region, yet in a modern microservices application a single request can trigger multiple interactions crossing an AZ boundary over and over again, through several network layers including Application Load Balancers and Kubernetes proxies. This can create an aggregated effect which is usually hard to detect and troubleshoot on both latency and performance, but also cost since data transfer charges apply for cross-AZ communication. Enhanced Berkeley Packet Filter (eBPF) offers unparalleled visibility into the network stack of a Kubernetes cluster. It can be used to unravel concealed performance bottlenecks and understand the nuanced cost implications of network requests cross AZs in Kubernetes. Click here to watch the replay!

Future of Intelligent Cluster Ops: LLM-Azing Kubernetes Controllers - Rajas Kakodkar, VMware & Amine Hilaly, AWS

As a Kubernetes operator, you must have spent countless hours upgrading clusters, deploying complex applications and troubleshooting issues. Have you ever wondered if you could automate this and literally speak to your cluster by asking - Is it safe to upgrade to v1.29? - Why isn’t Node X Ready? Join this session by Rajas and Amine to discover how AI can empower cluster operations with K8s controllers backed by LLMs. Discover the stages of data processing, fine-tuning LLMs and integrating them with K8s controllers and CRDs. And witness the addition of Speech Recognition to the K8s controller to operate clusters. To unravel the myths of AI hype, there will be a live demo to “talk” to K8s controller powered by LLM for - Auditing and upgrading clusters - Simulating Chaos scenarios - Scanning clusters for CVEs - Observability of cluster health The audience will get to know how domain knowledge helps improve AI model accuracy to ensure that it follows data ethics and security principles. Click here to watch the replay!

Intro + Deep Dive: Kubernetes SIG Scalability - Wojciech Tyczyński, Google & Shyam Jeedigunta, Amazon Web Services

This session will cover different efforts that SIG Scalability is involved in: defining what scalability means for Kubernetes, driving performance improvements, maintaining infrastructure for scalability testing, guarding Kubernetes against performance regressions. In addition to overall overview, the most recent achievement and challenges are always the top focus for the presentation. Cooperation with other SIGs is an important aspect of the presentation as many improvements driven from the SIG are in fact owned by other SIGs. Time for Q&A will be reserved at the end of the session to understand how the SIG can better engage with the community as well as to allow the audience to provide the input about the roadmap. Click here to watch the replay!

Zonal Outage Operational Stories - Jyoti Ranjan Mahapatra & Shyam Jeedigunta, Amazon Web Services

Most datacenters have a notion of “availability zone” as a failure domain. Correlated failures are expected in a single failure domain. Kubernetes cluster administrators deploy Kubernetes control plane, worker nodes, and pods, in a topological spread that can tolerate a single fault domain failure. Such setups achieve high availability and gracefully handle common zonal failures — network partitions, power-loss, reboot, bad software deployments, and so forth. This talk walks through numerous real world zonal outages, from a spectrum of partial to full outage, and the behavior of Kubernetes components in those situations. The speakers operate a large fleet of Kubernetes control plane in Amazon Web Services; they will share stories of zonal outages and improvements that helped achieve greater resiliency for thousands of clusters. Click here to watch the replay!

Kubernetes Maintainers Read Mean Comments - Tim Hockin, Google & Davanum Srinivas, Amazon Web Services

Being a maintainer of a large open-source project can sometimes be a thankless job. While most of our users are wonderful, sometimes things get heated, and occasionally people say something that just goes too far. Thankfully, we maintainers have each other to lean on, and good senses of humor. This session is a reminder that maintainers are just normal people, often doing this work out of passion. We share these nuggets to vent a little and to poke fun at ourselves as well. We love our users and the community who use our work. They are the most creative people and often build things we did not even think possible, but sometimes it gets to be a little too much ... and this is our escape valve. Click here to watch the replay!

Kubernetes SIG Architecture Intro and Updates - John Belamaric, Google & Davanum Srinivas, AWS

SIG Architecture maintains and evolves the design principles of Kubernetes, and provides a consistent body of expertise necessary to ensure architectural consistency over time. The SIG takes care of evolution of conformance definitions, API definitions/conventions, deprecation policy, design principles, and other cross-cutting concerns. In this talk, we will provide an introduction to SIG architecture, including its role and the various subprojects that support its activities. Additionally, we will provide a community update on the status of those efforts. Click here to watch the replay!

SIG Autoscaling Updates and Feature Highlights - Guy Templeton, Skyscanner; Jonathon Innis, AWS; Maciek Pytel, Google

Since adoption by SIG Autoscaling in the lead-up to Kubecon North America 2023, Karpenter has continued to develop its roadmap and integrations, allowing even more cluster operators to make use of it. Come hear the latest on the new features we've delivered and what we're planning for the future. If you're interested in the future of the project, want to get involved yourself and help move the project forward, or just have feedback on your experience, come along! Click here to watch the replay!

Keynote: Cloud Native in its Next Decade - Davanum Srinivas, Principal Engineer, AWS & Lin Sun, Head of Open Source, solo.io

When we started CNCF in 2015 to help advance container technology, Kubernetes was the seeding technology to provide a de facto container orchestration platform for all cloud native applications. Almost a decade later, the community has exploded with 180+ open source projects building on top of cloud native technologies. Looking ahead, what challenges will we have in the next decade? They will be vastly different for our users and contributors from today. Let us review some of the key CNCF projects today and lay out some possible avenues for where cloud native is going for the next decade, AI, sustainability, edge computing, security, service mesh, web assembly and more. Right or wrong, we’ll find out at KubeCon 2034! Click here to watch the replay!