KubeCon + CloudNativeCon 2024 Day Two
This day was highlighted the symbiotic relationship between Kubernetes and AI, emphasizing the need for standardization, efficiency, and collaboration to drive innovation in the cloud-native ecosystem.
Opening Keynote
Priyanka Sharma highlighted the significance of Kubernetes turning 10 years old and its pivotal role in the cloud-native landscape, particularly in the realm of AI. Cloud native technologies are crucial for AI innovation, especially in transitioning from prototyping to production. The need for AI standards and the release of a Cloud Native whitepaper on AI were emphasized.
Sharma demoed loading a Kubernetes cluster, taking a picture of the audience, and getting an AI-generated description of the photo immediately to demonstrate prototype to production at scale. Also, a new Cloud Native whitepaper on AI was just released, more proof the cloud native community is hard at work solving infrastructure problems for AI.
Accelerating AI Workloads with GPUs in Kubernetes
Kevin Klues and Sanjay Chatterjee discussed the importance of GPU acceleration in Kubernetes for AI workloads. Techniques for GPU support, dynamic resource allocation, and challenges such as topology-aware placement and fault tolerance were addressed. They called for collaborative efforts to solve complex problems using GPUs, AI, and Kubernetes.
Optimizing Performance and Sustainability for AI
The panel discussed strategies to enhance efficiency and sustainability of AI workloads on Kubernetes, emphasizing the importance of resource allocation and community collaboration. Techniques like attaching CPUs to GPU clusters and selecting specialized compute for AI models were highlighted.
Platform Building Blocks: How to Build ML Infrastructure with CNCF Projects
Yuzhui Liu and Leon Zhou presented Bloomberg's use case of crafting scalable ML infrastructure using cloud-native ecosystem tools. They showcased examples of AI applications in data extraction, enrichment, search, and summarization, enhancing Bloomberg terminal functions.
Kubernetes Pattern Evolution
Bilgin Ibryam and Roland Huß discussed the evolution of Kubernetes patterns and presented reusable elements for designing cloud-native applications. The presentation was based on their book "Kubernetes Patterns", available for download here with a free Red Hat Developer account.
AI Hub is Back!:
Rajarajan Pudupatti shared insights from Fidelity's work in leveraging AI, focusing on creating internal developer platforms for AI innovation. Techniques for improving user experience with AI, leveraging Kubernetes, and overcoming production complexities were discussed.